Merge commit '204af77596345c120e25ce9d433dad0676a59b37'

Record why the G005 branch-recovery work satisfies the roadmap pinpoints without touching leader-owned Ultragoal state.

Constraint: Task 2 requested ROADMAP.md/plan pinpoint mapping and explicitly forbids .omx/ultragoal mutation.

Rejected: leader-only mailbox note | the task prefers a repo-local docs/g005 verification map when unclaimed and absent.

Confidence: high

Scope-risk: narrow

Directive: Keep this map evidence-only; do not treat it as a substitute for leader Ultragoal checkpoints.

Tested: documentation-only map cross-checked against ROADMAP.md, prd.json, and task-1 verification output.

Not-tested: no code tests rerun after documentation-only commit.

.omx/cc2/issue-parity-intake.json

@@ -0,0 +1,429 @@
+{
+  "schema_version": "cc2.issue_parity_intake.v1",
+  "generated_at": "2026-05-14T08:02:00Z",
+  "task_id": "3",
+  "owner": "worker-2",
+  "goal": "G001-stream0-board",
+  "notes": [
+    "Leader owns Ultragoal; this artifact does not mutate .omx/ultragoal.",
+    "Rows are scoped intake/classification evidence for Worker 1/Task 2 board integration."
+  ],
+  "source_manifest": {
+    "claw_open_latest": {
+      "path": ".omx/research/claw-open-latest.json",
+      "sha256_prefix_from_plan": "89e3e027fa735f38",
+      "covered_issue_numbers": [3028, 3029, 3030, 3031, 3032, 3033, 3034, 3035, 3036, 3037, 3038]
+    },
+    "claw_issues": {
+      "path": ".omx/research/claw-issues.json",
+      "sha256_prefix_from_plan": "e64fdba7df3b78ed",
+      "covered_issue_numbers": [2997, 3003, 3004, 3005, 3006, 3007, 3020, 3023]
+    },
+    "opencode": {
+      "repo_path": ".omx/research/repos/opencode",
+      "metadata_path": ".omx/research/opencode-repo.json",
+      "issues_path": ".omx/research/opencode-issues.json",
+      "head_from_plan": "27ac53aaacc677b1401c4e75ca7a7dadf8b2c349"
+    },
+    "codex": {
+      "repo_path": ".omx/research/repos/codex",
+      "metadata_path": ".omx/research/codex-repo.json",
+      "issues_path": ".omx/research/codex-issues.json",
+      "head_from_plan": "6a225e4005209f2325ab3c681c7c6beba2907d4d"
+    }
+  },
+  "issue_clusters": [
+    {
+      "id": "CC2-ISSUE-3007",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3007",
+      "source_type": "github_issue",
+      "source_number": 3007,
+      "title": "Permission modes do not enforce path scope on file tools or shell expansion in bash",
+      "theme": "security/path-scope",
+      "release_bucket": "alpha_blocker",
+      "lifecycle_status": "active",
+      "roadmap_anchor": "ROADMAP.md#11-policy-engine-for-autonomous-coding; ROADMAP.md#9-green-ness-contract",
+      "dependencies": ["permission path canonicalization", "file tool target validation", "bash command/path validation reachability", "policy regression fixtures"],
+      "verification_required": ["workspace-write cannot read/write/delete outside workspace", "shell expansion and symlink traversal are rejected or policy-blocked", "file tools and bash use the same target-scope decision record"],
+      "deferral_rationale": null,
+      "classification_rationale": "Security/sandbox escape class; plan names #3007 as alpha blocker."
+    },
+    {
+      "id": "CC2-ISSUE-3020",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3020",
+      "source_type": "github_issue",
+      "source_number": 3020,
+      "title": "OpenAI-compatible model IDs with slashes are stripped before request",
+      "theme": "provider/model-routing",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#provider-routing-model-name-prefix-must-win-over-env-var-presence-fixed-2026-04-08-0530c50",
+      "dependencies": ["provider profile contract", "wire model-id preservation option", "routing-prefix source reporting"],
+      "verification_required": ["OpenAI-compatible endpoint receives exact model id when preservation is enabled", "status JSON reports raw model input, route, and wire model id"],
+      "deferral_rationale": null,
+      "classification_rationale": "Core provider correctness but below alpha state/security contracts unless it blocks the selected alpha model path."
+    },
+    {
+      "id": "CC2-ISSUE-3006",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3006",
+      "source_type": "github_issue",
+      "source_number": 3006,
+      "title": "Not Working in windows",
+      "theme": "windows/install",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#immediate-backlog-from-current-real-pain",
+      "dependencies": ["Windows support policy", "PowerShell install path", "dependency/version matrix", "diagnostic setup output"],
+      "verification_required": ["fresh Windows/PowerShell setup smoke documented", "unsupported native paths fail with actionable WSL2/native guidance"],
+      "deferral_rationale": null,
+      "classification_rationale": "Real adoption blocker; plan places Windows/install in beta adoption overlay."
+    },
+    {
+      "id": "CC2-ISSUE-3005",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3005",
+      "source_type": "github_issue",
+      "source_number": 3005,
+      "title": "DeepSeek V4-flash/pro fails with 400 Bad Request (missing reasoning_content) while deepseek-reasoner works",
+      "theme": "provider/response-shape",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#5-failure-taxonomy; ROADMAP.md#provider-routing-model-name-prefix-must-win-over-env-var-presence-fixed-2026-04-08-0530c50",
+      "dependencies": ["OpenAI-compatible diagnostics playbook", "provider error taxonomy", "reasoning/thinking field compatibility tests"],
+      "verification_required": ["provider 400 response classified with actionable remediation", "DeepSeek-compatible response-shape fixture does not hide assistant output"],
+      "deferral_rationale": null,
+      "classification_rationale": "Provider compatibility issue that shares the #3032 diagnostics lane."
+    },
+    {
+      "id": "CC2-ISSUE-3004",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3004",
+      "source_type": "github_issue",
+      "source_number": 3004,
+      "title": "When can we adapt to zed?",
+      "theme": "ide/acp",
+      "release_bucket": "ga_ecosystem",
+      "lifecycle_status": "deferred_with_rationale",
+      "roadmap_anchor": "ROADMAP.md#phase-5-plugin-and-mcp-lifecycle-maturity",
+      "dependencies": ["stable session/control API", "plugin/MCP lifecycle", "engine API or ACP bridge decision"],
+      "verification_required": ["Zed/ACP smoke once core state/control contracts exist"],
+      "deferral_rationale": "IDE integration is valuable but should wait until boot/session/event/control truth surfaces are stable.",
+      "classification_rationale": "Matches plan's GA ecosystem lane for Zed/ACP."
+    },
+    {
+      "id": "CC2-ISSUE-3003",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3003",
+      "source_type": "github_issue",
+      "source_number": 3003,
+      "title": ".claude/sessions should not be submitted to repo",
+      "theme": "session-hygiene/gitignore",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#9-green-ness-contract; ROADMAP.md#8-recovery-recipes-for-common-failures",
+      "dependencies": ["artifact ignore policy", "session storage boundary docs", "repo hygiene check"],
+      "verification_required": ["session directories are ignored", "status/doctor warns about tracked session artifacts"],
+      "deferral_rationale": null,
+      "classification_rationale": "Small but user-visible session hygiene and data-leak prevention item."
+    },
+    {
+      "id": "CC2-ISSUE-2997",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/2997",
+      "source_type": "github_issue",
+      "source_number": 2997,
+      "title": "License?",
+      "theme": "docs/license",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#immediate-backlog-from-current-real-pain",
+      "dependencies": ["maintainer license decision", "LICENSE file", "README/USAGE attribution wording"],
+      "verification_required": ["repository license file exists", "package metadata and docs reference the same license"],
+      "deferral_rationale": null,
+      "classification_rationale": "Adoption/readiness documentation gap; requires maintainer decision before implementation."
+    },
+    {
+      "id": "CC2-ISSUE-3023",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3023",
+      "source_type": "github_issue",
+      "source_number": 3023,
+      "title": "Protect claw-code from AI slop PRs",
+      "theme": "repo-hygiene/anti-slop",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#immediate-backlog-from-current-real-pain",
+      "dependencies": ["contributor policy", "PR quality gate selection", "false-positive review escape hatch"],
+      "verification_required": ["selected PR quality gate runs on sample good/bad PR fixtures", "maintainers can override false positives"],
+      "deferral_rationale": null,
+      "classification_rationale": "Protects project throughput but should not precede alpha core safety contracts."
+    },
+    {
+      "id": "CC2-ISSUE-3028",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3028",
+      "source_type": "github_issue",
+      "source_number": 3028,
+      "title": "docs: add navigation and file-context usage guide",
+      "theme": "docs/navigation-context",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#7-human-ux-still-leaks-into-claw-workflows",
+      "dependencies": ["current TUI/shell key behavior inventory", "file context syntax docs", "secret-handling guidance"],
+      "verification_required": ["docs include terminal history, scrollback, @file context, attach/external file caveats", "examples work against current CLI"],
+      "deferral_rationale": null,
+      "classification_rationale": "Documentation support item from latest open issue refresh."
+    },
+    {
+      "id": "CC2-ISSUE-3029",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3029",
+      "source_type": "github_issue",
+      "source_number": 3029,
+      "title": "build: add cross-platform installer path and release artifact quickstart",
+      "theme": "install/distribution",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#immediate-backlog-from-current-real-pain",
+      "dependencies": ["release artifact policy", "install.sh/install.ps1 contract", "PATH/update/uninstall instructions"],
+      "verification_required": ["install quickstart smoke on supported OS/arch", "failed install prints actionable diagnostics"],
+      "deferral_rationale": null,
+      "classification_rationale": "Distribution friction belongs in adoption overlay."
+    },
+    {
+      "id": "CC2-ISSUE-3030",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3030",
+      "source_type": "github_issue",
+      "source_number": 3030,
+      "title": "feat: make provider/model setup less env-var-driven",
+      "theme": "provider/setup-profiles",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#3-structured-session-control-api; ROADMAP.md#145-boot-preflight-doctor-contract",
+      "dependencies": ["provider profiles", "setup wizard or dry-run", "secret redaction", "base-url/model smoke test"],
+      "verification_required": ["setup validates provider route without echoing keys", "session-only versus persisted profile behavior is explicit"],
+      "deferral_rationale": null,
+      "classification_rationale": "Directly reduces current provider setup support churn."
+    },
+    {
+      "id": "CC2-ISSUE-3031",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3031",
+      "source_type": "github_issue",
+      "source_number": 3031,
+      "title": "feat: auto-compact or clearly recover from context-window provider errors",
+      "theme": "session-recovery/context-window",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#8-recovery-recipes-for-common-failures; ROADMAP.md#158-compact_messages_if_needed-drops-turns-silently-no-structured-compaction-event-emitted",
+      "dependencies": ["provider error classifier", "safe compact retry policy", "compaction event/audit trail", "retry loop cap"],
+      "verification_required": ["context-window error either compacts+retries once safely or emits exact recovery command", "compaction event is machine-visible"],
+      "deferral_rationale": null,
+      "classification_rationale": "Recovery reliability item; promoted only if selected alpha provider path hits it."
+    },
+    {
+      "id": "CC2-ISSUE-3032",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3032",
+      "source_type": "github_issue",
+      "source_number": 3032,
+      "title": "docs: add OpenAI-compatible/local provider diagnostics playbook",
+      "theme": "provider/diagnostics-docs",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#5-failure-taxonomy",
+      "dependencies": ["raw chat-completions smoke tests", "tool-call response-shape examples", "provider failure taxonomy"],
+      "verification_required": ["playbook distinguishes Claw bugs from wrapper/tool-call-shape bugs", "curl examples cover non-streaming and streaming tool calls"],
+      "deferral_rationale": null,
+      "classification_rationale": "Shared diagnostic lane for #3005/#3020/local model reports."
+    },
+    {
+      "id": "CC2-ISSUE-3033",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3033",
+      "source_type": "github_issue",
+      "source_number": 3033,
+      "title": "feat: add minimal claw serve JSON-RPC engine API",
+      "theme": "engine-api/control-plane",
+      "release_bucket": "ga_ecosystem",
+      "lifecycle_status": "deferred_with_rationale",
+      "roadmap_anchor": "ROADMAP.md#3-structured-session-control-api; ROADMAP.md#phase-4-claws-first-task-execution",
+      "dependencies": ["stable session state API", "event schema v1", "permission policy contract", "cancel/prompt stream semantics"],
+      "verification_required": ["protocol conformance fixtures for session/create prompt/stream cancel error", "capability negotiation backwards compatibility"],
+      "deferral_rationale": "Engine API should expose, not invent, stable core control-plane semantics after alpha contracts land.",
+      "classification_rationale": "Useful integration surface but too broad for alpha unless narrowed to existing session control API."
+    },
+    {
+      "id": "CC2-ISSUE-3034",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3034",
+      "source_type": "github_issue",
+      "source_number": 3034,
+      "title": "docs: define evidence-gated Hermes handoff loop for Claw Code execution",
+      "theme": "sdlc/evidence-handoff",
+      "release_bucket": "post_2_0_research",
+      "lifecycle_status": "deferred_with_rationale",
+      "roadmap_anchor": "ROADMAP.md#4-canonical-lane-event-schema; ROADMAP.md#10-typed-task-packet-format",
+      "dependencies": ["typed task packet", "evidence bundle schema", "report gate status vocabulary"],
+      "verification_required": ["handoff packet fixture validates scope/success/test evidence fields", "post-flight gate consumes evidence instead of free-text summary"],
+      "deferral_rationale": "Can inform event/report/task contracts, but Hermes-specific loop should stay research/docs until core schemas are stable.",
+      "classification_rationale": "Only the generic evidence-gated contract is Claw 2.0; Hermes branding is not core."
+    },
+    {
+      "id": "CC2-ISSUE-3035",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3035",
+      "source_type": "github_issue",
+      "source_number": 3035,
+      "title": "fix: improve compacted session resume discoverability",
+      "theme": "session-resume/discoverability",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#8-recovery-recipes-for-common-failures; ROADMAP.md#160-session_store-has-no-list_sessions-delete_session-or-session_exists",
+      "dependencies": ["session enumeration", "latest-session workspace search boundary", "compacted session marker"],
+      "verification_required": ["/resume latest finds newest eligible compacted session", "/session or status lists resumable compacted sessions with path/id"],
+      "deferral_rationale": null,
+      "classification_rationale": "Session recovery/adoption item."
+    },
+    {
+      "id": "CC2-ISSUE-3036",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3036",
+      "source_type": "github_issue",
+      "source_number": 3036,
+      "title": "docs: add official Ollama/llama.cpp/vLLM local model examples",
+      "theme": "provider/local-docs",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#145-boot-preflight-doctor-contract; ROADMAP.md#5-failure-taxonomy",
+      "dependencies": ["known-good local provider examples", "raw /v1 smoke test", "tool-call limitation warning"],
+      "verification_required": ["docs include Ollama/llama.cpp/vLLM examples and HELLO smoke", "tool-call caveats are explicit"],
+      "deferral_rationale": null,
+      "classification_rationale": "Local provider adoption support."
+    },
+    {
+      "id": "CC2-ISSUE-3037",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3037",
+      "source_type": "github_issue",
+      "source_number": 3037,
+      "title": "docs: clarify Claw Code positioning as multi-provider Claude-Code-shaped runtime",
+      "theme": "docs/product-positioning",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "roadmap_anchor": "ROADMAP.md#goal; ROADMAP.md#definition-of-clawable",
+      "dependencies": ["README positioning copy", "provider support truth table", "identity leak bug policy"],
+      "verification_required": ["README/docs answer Claude-only question directly", "provider support wording matches implemented routes"],
+      "deferral_rationale": null,
+      "classification_rationale": "Clarifies product identity for adoption without broad implementation."
+    },
+    {
+      "id": "CC2-ISSUE-3038",
+      "source_anchor": "https://github.com/ultraworkers/claw-code/issues/3038",
+      "source_type": "github_issue",
+      "source_number": 3038,
+      "title": "roadmap: track skills/plugins/marketplace ecosystem gap after core UX stabilizes",
+      "theme": "plugin-marketplace/ecosystem",
+      "release_bucket": "ga_ecosystem",
+      "lifecycle_status": "deferred_with_rationale",
+      "roadmap_anchor": "ROADMAP.md#13-first-class-pluginmcp-lifecycle-contract; ROADMAP.md#14-mcp-end-to-end-lifecycle-parity",
+      "dependencies": ["plugin/MCP lifecycle contract", "extension point inventory", "discovery/install/update flow design"],
+      "verification_required": ["extension point inventory exists", "marketplace work explicitly depends on core UX stabilization"],
+      "deferral_rationale": "Marketplace breadth should wait until core setup/auth/provider/session UX and plugin lifecycle are reliable.",
+      "classification_rationale": "Matches plan's ga_ecosystem/post-2.0 caution for marketplace parity."
+    }
+  ],
+  "parity_rows": [
+    {
+      "id": "CC2-PARITY-OPENCODE-PLUGIN-ECOSYSTEM",
+      "source_anchor": "anomalyco/opencode@27ac53aa packages/app/web/desktop/plugin/sdk/extensions/zed/slack/containers plus issue #3038",
+      "source_type": "repo_clone_and_local_issue",
+      "title": "Plugin/skills/marketplace ecosystem inventory",
+      "release_bucket": "ga_ecosystem",
+      "lifecycle_status": "deferred_with_rationale",
+      "dependencies": ["Claw plugin/MCP lifecycle contract", "current extension-point inventory"],
+      "verification_required": ["inventory maps current Claw plugin/skill/MCP extension points before marketplace implementation"],
+      "deferral_rationale": "Adapt ecosystem discovery only after core setup/provider/session reliability is stable."
+    },
+    {
+      "id": "CC2-PARITY-OPENCODE-PERMISSION-PRESETS",
+      "source_anchor": "https://github.com/anomalyco/opencode/issues/27464 and ROADMAP.md#11-policy-engine-for-autonomous-coding",
+      "source_type": "external_issue_and_roadmap",
+      "title": "Quick permission preset switching mapped onto Claw policy profiles",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "dependencies": ["policy profile model", "approval-token audit trail"],
+      "verification_required": ["preset switch is visible in status/report output and cannot bypass path-scope enforcement"],
+      "deferral_rationale": null
+    },
+    {
+      "id": "CC2-PARITY-OPENCODE-CUSTOM-PROVIDER-PARAMS",
+      "source_anchor": "https://github.com/anomalyco/opencode/issues/27462 and #3030/#3032",
+      "source_type": "external_issue_and_local_issue",
+      "title": "Custom API parameter passthrough for provider profiles",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "dependencies": ["provider profile schema", "secret redaction", "request audit surface"],
+      "verification_required": ["custom params are schema-validated, redacted, and visible as provenance without leaking secrets"],
+      "deferral_rationale": null
+    },
+    {
+      "id": "CC2-PARITY-OPENCODE-TODOWRITE-AUTOCOMPLETE",
+      "source_anchor": "https://github.com/anomalyco/opencode/issues/27453 and ROADMAP.md#10-typed-task-packet-format",
+      "source_type": "external_issue_and_roadmap",
+      "title": "Task/Todo completion assistance via typed task lifecycle",
+      "release_bucket": "ga_ecosystem",
+      "lifecycle_status": "deferred_with_rationale",
+      "dependencies": ["typed task packet", "task lifecycle events", "evidence-gated completion"],
+      "verification_required": ["auto-complete suggestions cannot mark work complete without evidence bundle or explicit user approval"],
+      "deferral_rationale": "Useful UX should follow, not precede, typed task lifecycle and evidence contract."
+    },
+    {
+      "id": "CC2-PARITY-OPENCODE-WINDOWS-DISTRIBUTION",
+      "source_anchor": "https://github.com/anomalyco/opencode/issues/27476 https://github.com/anomalyco/opencode/issues/27459 https://github.com/anomalyco/opencode/issues/27470 and #3006/#3029",
+      "source_type": "external_issues_and_local_issues",
+      "title": "Windows/GLIBC/distribution reliability parity lessons",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "dependencies": ["install artifact matrix", "Windows encoding guidance", "minimum Linux/GLIBC support statement"],
+      "verification_required": ["release quickstart documents supported OS matrix and known terminal/encoding caveats"],
+      "deferral_rationale": null
+    },
+    {
+      "id": "CC2-PARITY-CODEX-GRANULAR-PERMISSIONS",
+      "source_anchor": "https://github.com/openai/codex/issues/22595 and Codex docs permissions/app/plugin concepts",
+      "source_type": "external_issue_and_docs",
+      "title": "Granular app/plugin permissions adapted to Claw policy engine",
+      "release_bucket": "alpha_blocker",
+      "lifecycle_status": "active",
+      "dependencies": ["permission enforcer path-scope fix", "plugin/MCP capability model", "approval-token replay protection"],
+      "verification_required": ["granular permission grants do not widen workspace path scope implicitly"],
+      "deferral_rationale": null
+    },
+    {
+      "id": "CC2-PARITY-CODEX-SESSION-RECOVERY",
+      "source_anchor": "https://github.com/openai/codex/issues/22619 https://github.com/openai/codex/issues/22597 https://github.com/openai/codex/issues/22593 and #3035",
+      "source_type": "external_issues_and_local_issue",
+      "title": "Safe local session/thread recovery without storage amplification",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "dependencies": ["session enumeration", "resume latest boundary", "JSONL/storage compaction policy"],
+      "verification_required": ["recoverable sessions are discoverable and session forks avoid unbounded duplicate history"],
+      "deferral_rationale": null
+    },
+    {
+      "id": "CC2-PARITY-CODEX-PROXY-NETWORK",
+      "source_anchor": "https://github.com/openai/codex/issues/22623 and #3032",
+      "source_type": "external_issue_and_local_issue",
+      "title": "Provider/network diagnostics include proxy behavior",
+      "release_bucket": "beta_adoption",
+      "lifecycle_status": "open",
+      "dependencies": ["HTTP client proxy detection", "provider diagnostics playbook"],
+      "verification_required": ["diagnostics report whether proxy env/config is honored for provider calls"],
+      "deferral_rationale": null
+    },
+    {
+      "id": "CC2-PARITY-CODEX-CLI-AGENT-FLAG",
+      "source_anchor": "https://github.com/openai/codex/issues/22615 and ROADMAP.md#10-typed-task-packet-format",
+      "source_type": "external_issue_and_roadmap",
+      "title": "CLI flag for agent/subagent mode mapped to Claw typed task packets",
+      "release_bucket": "ga_ecosystem",
+      "lifecycle_status": "deferred_with_rationale",
+      "dependencies": ["typed task packet", "session control API", "policy-scoped worker launch"],
+      "verification_required": ["CLI agent mode cannot bypass task policy or evidence requirements"],
+      "deferral_rationale": "Implement only after core task/session control contracts are stable."
+    }
+  ],
+  "coverage": {
+    "required_latest_open_range_3028_3038": [3028, 3029, 3030, 3031, 3032, 3033, 3034, 3035, 3036, 3037, 3038],
+    "required_existing_issue_numbers": [3007, 3006, 3020, 3005, 3003, 2997, 3023, 3004],
+    "issue_rows_expected": 19,
+    "parity_rows_expected_minimum": 6
+  }
+}

.omx/cc2/issue-parity-intake.md

@@ -0,0 +1,47 @@
+# CC2 Issue / Parity Intake Mapping
+
+Generated by `worker-2` for team task 3 (`G001 issue/parity intake mapping`). This is a board-integration fragment for Stream 0; it intentionally does **not** mutate `.omx/ultragoal`.
+
+## Covered local issue clusters
+
+| Issue | Theme | Bucket | Lifecycle | Board anchor |
+|---:|---|---|---|---|
+| #3007 | security/path-scope | `alpha_blocker` | `active` | Policy engine + green-ness contract |
+| #3020 | provider/model-routing | `beta_adoption` | `open` | Provider routing/model source status |
+| #3006 | windows/install | `beta_adoption` | `open` | Immediate backlog / install readiness |
+| #3005 | provider/response-shape | `beta_adoption` | `open` | Failure taxonomy / provider diagnostics |
+| #3004 | ide/acp | `ga_ecosystem` | `deferred_with_rationale` | Plugin/MCP lifecycle maturity |
+| #3003 | session-hygiene/gitignore | `beta_adoption` | `open` | Green-ness / recovery hygiene |
+| #2997 | docs/license | `beta_adoption` | `open` | Adoption docs/license readiness |
+| #3023 | repo-hygiene/anti-slop | `beta_adoption` | `open` | Immediate backlog / PR quality gate |
+| #3028 | docs/navigation-context | `beta_adoption` | `open` | Human UX leaks into claw workflows |
+| #3029 | install/distribution | `beta_adoption` | `open` | Cross-platform release quickstart |
+| #3030 | provider/setup-profiles | `beta_adoption` | `open` | Boot preflight / structured session control |
+| #3031 | session-recovery/context-window | `beta_adoption` | `open` | Recovery recipes / compaction event |
+| #3032 | provider/diagnostics-docs | `beta_adoption` | `open` | Failure taxonomy |
+| #3033 | engine-api/control-plane | `ga_ecosystem` | `deferred_with_rationale` | Structured session control API |
+| #3034 | sdlc/evidence-handoff | `post_2_0_research` | `deferred_with_rationale` | Event/report/task contract input |
+| #3035 | session-resume/discoverability | `beta_adoption` | `open` | Recovery recipes / session enumeration |
+| #3036 | provider/local-docs | `beta_adoption` | `open` | Provider setup and diagnostics docs |
+| #3037 | docs/product-positioning | `beta_adoption` | `open` | Goal / definition of clawable |
+| #3038 | plugin-marketplace/ecosystem | `ga_ecosystem` | `deferred_with_rationale` | Plugin/MCP lifecycle maturity |
+
+## Parity intake rows
+
+| Row | Source | Bucket | Lifecycle | Adaptation rule |
+|---|---|---|---|---|
+| `CC2-PARITY-OPENCODE-PLUGIN-ECOSYSTEM` | opencode repo + #3038 | `ga_ecosystem` | `deferred_with_rationale` | Inventory Claw extension points before marketplace work. |
+| `CC2-PARITY-OPENCODE-PERMISSION-PRESETS` | opencode #27464 | `beta_adoption` | `open` | Permission preset UX must not bypass Claw path-scope policy. |
+| `CC2-PARITY-OPENCODE-CUSTOM-PROVIDER-PARAMS` | opencode #27462 + #3030/#3032 | `beta_adoption` | `open` | Custom provider params need schema validation, redaction, and provenance. |
+| `CC2-PARITY-OPENCODE-TODOWRITE-AUTOCOMPLETE` | opencode #27453 | `ga_ecosystem` | `deferred_with_rationale` | Auto-complete task UX follows typed task lifecycle/evidence gates. |
+| `CC2-PARITY-OPENCODE-WINDOWS-DISTRIBUTION` | opencode #27476/#27459/#27470 + #3006/#3029 | `beta_adoption` | `open` | Use external pain as release-matrix and diagnostics evidence. |
+| `CC2-PARITY-CODEX-GRANULAR-PERMISSIONS` | Codex #22595 + docs | `alpha_blocker` | `active` | Adapt granular permissions only through Claw policy engine and approval tokens. |
+| `CC2-PARITY-CODEX-SESSION-RECOVERY` | Codex #22619/#22597/#22593 + #3035 | `beta_adoption` | `open` | Session discovery/recovery must avoid storage amplification. |
+| `CC2-PARITY-CODEX-PROXY-NETWORK` | Codex #22623 + #3032 | `beta_adoption` | `open` | Provider diagnostics should expose proxy behavior. |
+| `CC2-PARITY-CODEX-CLI-AGENT-FLAG` | Codex #22615 | `ga_ecosystem` | `deferred_with_rationale` | CLI agent mode waits for typed task/session control contracts. |
+
+Validation command:
+
+```bash
+python3 .omx/cc2/validate_issue_parity_intake.py
+```

.omx/cc2/render_board_md.py

@@ -0,0 +1,250 @@
+#!/usr/bin/env python3
+"""Render the Claw Code 2.0 canonical board JSON as a human-readable Markdown board."""
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from collections import Counter, defaultdict
+from pathlib import Path
+from typing import Any
+
+STATUS_DESCRIPTIONS = {
+    "context": "Context-only heading or evidence anchor; not an implementation work item.",
+    "active": "Current Claw Code 2.0 implementation surface that should remain visible on the board.",
+    "open": "Actionable unresolved work that needs implementation or acceptance evidence.",
+    "done_verify": "Marked as done upstream but retained for verification against current CC2 behavior.",
+    "stale_done": "Historically completed or merged work that may be stale and needs freshness checks before relying on it.",
+    "superseded": "Replaced by a newer item; keep as traceability context only.",
+    "deferred_with_rationale": "Intentionally deferred; rationale must be present in the board item.",
+    "rejected_not_claw": "Excluded because it is not Claw Code product work.",
+}
+
+BUCKET_DESCRIPTIONS = {
+    "alpha_blocker": "Must be resolved before alpha-quality autonomous coding lanes are dependable.",
+    "beta_adoption": "Important for broader dogfood/adoption once alpha blockers are controlled.",
+    "ga_ecosystem": "Required for mature plugin/MCP/provider ecosystem behavior.",
+    "2.x_intake": "Post-2.0 intake or follow-up candidate retained for sequencing.",
+    "post_2_0_research": "Research-oriented item not required for the CC2 board cut.",
+    "context": "Non-actionable roadmap context.",
+    "rejected_not_claw": "Explicit non-Claw rejection bucket.",
+}
+
+LANE_TITLES = {
+    "stream_0_governance": "Stream 0 — Governance, intake, and cross-cutting roadmap triage",
+    "stream_1_worker_boot_session_control": "Stream 1 — Worker boot and session control",
+    "stream_2_event_reporting_contracts": "Stream 2 — Event/reporting contracts",
+    "stream_3_branch_test_recovery": "Stream 3 — Branch/test recovery",
+    "stream_4_claws_first_execution": "Stream 4 — Claws-first task execution",
+    "stream_5_plugin_mcp_lifecycle": "Stream 5 — Plugin/MCP lifecycle",
+    "adoption_overlay": "Adoption overlay — user-visible parity and release polish",
+    "parity_overlay": "Parity overlay — opencode/codex comparison context",
+}
+
+REQUIRED_ITEM_FIELDS = [
+    "id",
+    "title",
+    "source_anchor",
+    "source_type",
+    "release_bucket",
+    "lifecycle_status",
+    "dependencies",
+    "verification_required",
+    "deferral_rationale",
+]
+
+
+def load_board(path: Path) -> dict[str, Any]:
+    with path.open() as f:
+        board = json.load(f)
+    if not isinstance(board, dict):
+        raise ValueError("board JSON root must be an object")
+    items = board.get("items")
+    if not isinstance(items, list):
+        raise ValueError("board JSON must contain an items array")
+    return board
+
+
+def validate_board(board: dict[str, Any]) -> list[str]:
+    errors: list[str] = []
+    coverage = board.get("coverage", {})
+    if coverage.get("unmapped_roadmap_heading_lines"):
+        errors.append(f"unmapped roadmap heading lines: {coverage['unmapped_roadmap_heading_lines']}")
+    if coverage.get("roadmap_headings_mapped") != coverage.get("roadmap_headings_total"):
+        errors.append("roadmap heading coverage is incomplete")
+    if coverage.get("roadmap_actions_mapped") != coverage.get("roadmap_actions_total"):
+        errors.append("roadmap ordered-action coverage is incomplete")
+
+    allowed_status = set(board.get("generation_policy", {}).get("status_values", []))
+    allowed_buckets = set(board.get("generation_policy", {}).get("release_buckets", []))
+    seen_ids: set[str] = set()
+    for index, item in enumerate(board["items"], 1):
+        for field in REQUIRED_ITEM_FIELDS:
+            if field not in item:
+                errors.append(f"item {index} missing required field {field}")
+        item_id = item.get("id")
+        if item_id in seen_ids:
+            errors.append(f"duplicate item id {item_id}")
+        seen_ids.add(item_id)
+        status = item.get("lifecycle_status")
+        bucket = item.get("release_bucket")
+        if allowed_status and status not in allowed_status:
+            errors.append(f"{item_id} has unknown lifecycle_status {status!r}")
+        if allowed_buckets and bucket not in allowed_buckets:
+            errors.append(f"{item_id} has unknown release_bucket {bucket!r}")
+        if status == "deferred_with_rationale" and not str(item.get("deferral_rationale", "")).strip():
+            errors.append(f"{item_id} is deferred without deferral_rationale")
+    return errors
+
+
+def table(headers: list[str], rows: list[list[Any]]) -> list[str]:
+    out = ["| " + " | ".join(headers) + " |", "| " + " | ".join("---" for _ in headers) + " |"]
+    for row in rows:
+        out.append("| " + " | ".join(str(cell) for cell in row) + " |")
+    return out
+
+
+def fmt_list(value: Any) -> str:
+    if not value:
+        return "none"
+    if isinstance(value, list):
+        return ", ".join(f"`{v}`" for v in value) if value else "none"
+    return f"`{value}`"
+
+
+def render(board: dict[str, Any]) -> str:
+    items: list[dict[str, Any]] = board["items"]
+    summary = board.get("summary", {})
+    coverage = board.get("coverage", {})
+    sources = board.get("sources", {})
+    policy = board.get("generation_policy", {})
+    by_lane = Counter(item.get("owner_lane", "unassigned") for item in items)
+    by_status = Counter(item.get("lifecycle_status", "unknown") for item in items)
+    by_bucket = Counter(item.get("release_bucket", "unknown") for item in items)
+    by_source = Counter(item.get("source_type", "unknown") for item in items)
+
+    lines: list[str] = []
+    lines.append("# Claw Code 2.0 Canonical Board")
+    lines.append("")
+    lines.append(f"Generated from board schema: `{board.get('generated_at', 'unknown')}`")
+    lines.append(f"Schema version: `{board.get('schema_version', 'unknown')}`")
+    lines.append("Ultragoal mutation policy: `.omx/ultragoal` is leader-owned and was not modified by this rendering task.")
+    lines.append("")
+
+    lines.append("## Evidence Freeze")
+    lines.append("")
+    roadmap = sources.get("roadmap", {})
+    research = sources.get("research", {})
+    plan = sources.get("approved_plan", {})
+    lines.extend(table(["Source", "Frozen evidence"], [
+        ["Roadmap", f"`{roadmap.get('path', 'ROADMAP.md')}` sha256 prefix `{roadmap.get('sha256_prefix', 'unknown')}`; {roadmap.get('heading_count', '?')} headings; {roadmap.get('ordered_action_count', '?')} ordered actions"],
+        ["Approved plan", f"`{plan.get('path', '.omx/plans/claw-code-2-0-adaptive-plan.md')}` sha256 prefix `{plan.get('sha256_prefix', 'unknown')}`"],
+        ["Research bundle", f"root `{research.get('root', '.omx/research')}`; latest open issues {research.get('claw_open_latest_count', '?')}; issue corpus {research.get('claw_issues_count', '?')}; codex/opencode clone metadata included"],
+    ]))
+    lines.append("")
+
+    lines.append("## Roadmap Coverage Summary")
+    lines.append("")
+    heading_total = coverage.get("roadmap_headings_total", 0)
+    heading_mapped = coverage.get("roadmap_headings_mapped", 0)
+    action_total = coverage.get("roadmap_actions_total", 0)
+    action_mapped = coverage.get("roadmap_actions_mapped", 0)
+    lines.extend(table(["Coverage gate", "Mapped", "Total", "Status"], [
+        ["ROADMAP headings", heading_mapped, heading_total, "PASS" if heading_mapped == heading_total and not coverage.get("unmapped_roadmap_heading_lines") else "FAIL"],
+        ["ROADMAP ordered actions", action_mapped, action_total, "PASS" if action_mapped == action_total else "FAIL"],
+        ["Duplicate heading lines", len(coverage.get("duplicate_roadmap_heading_lines", [])), 0, "PASS" if not coverage.get("duplicate_roadmap_heading_lines") else "WARN"],
+    ]))
+    lines.append("")
+    lines.append(f"Total canonical board items: **{len(items)}**")
+    lines.append("")
+
+    lines.append("## Lifecycle Enum Reference")
+    lines.append("")
+    status_rows = []
+    for status in policy.get("status_values", sorted(by_status)):
+        status_rows.append([f"`{status}`", by_status.get(status, 0), STATUS_DESCRIPTIONS.get(status, "Board-defined lifecycle status.")])
+    lines.extend(table(["Lifecycle", "Count", "Meaning"], status_rows))
+    lines.append("")
+
+    lines.append("## Release Bucket Reference")
+    lines.append("")
+    bucket_rows = []
+    for bucket in policy.get("release_buckets", sorted(by_bucket)):
+        bucket_rows.append([f"`{bucket}`", by_bucket.get(bucket, 0), BUCKET_DESCRIPTIONS.get(bucket, "Board-defined release bucket.")])
+    lines.extend(table(["Bucket", "Count", "Meaning"], bucket_rows))
+    lines.append("")
+
+    lines.append("## Stream Summaries")
+    lines.append("")
+    lane_rows = []
+    for lane, count in sorted(by_lane.items()):
+        lane_items = [item for item in items if item.get("owner_lane") == lane]
+        lane_status = Counter(item.get("lifecycle_status") for item in lane_items)
+        open_like = lane_status.get("active", 0) + lane_status.get("open", 0) + lane_status.get("done_verify", 0)
+        lane_rows.append([
+            LANE_TITLES.get(lane, lane),
+            count,
+            open_like,
+            ", ".join(f"`{k}` {v}" for k, v in sorted(lane_status.items())),
+        ])
+    lines.extend(table(["Stream / lane", "Items", "Active+open+verify", "Lifecycle mix"], lane_rows))
+    lines.append("")
+
+    lines.append("## Source-Type Mix")
+    lines.append("")
+    lines.extend(table(["Source type", "Items"], [[f"`{k}`", v] for k, v in sorted(by_source.items())]))
+    lines.append("")
+
+    lines.append("## Board Items by Stream")
+    lines.append("")
+    for lane in sorted(by_lane):
+        lane_items = [item for item in items if item.get("owner_lane") == lane]
+        lines.append(f"### {LANE_TITLES.get(lane, lane)}")
+        lines.append("")
+        lines.extend(table(
+            ["ID", "Title", "Source", "Bucket", "Lifecycle", "Verification", "Dependencies", "Deferral"],
+            [[
+                f"`{item.get('id')}`",
+                str(item.get("title", "")).replace("|", "\\|"),
+                f"`{item.get('source_anchor')}` / `{item.get('source_type')}`",
+                f"`{item.get('release_bucket')}`",
+                f"`{item.get('lifecycle_status')}`",
+                f"`{item.get('verification_required')}`",
+                fmt_list(item.get("dependencies")),
+                str(item.get("deferral_rationale") or "—").replace("|", "\\|"),
+            ] for item in lane_items]
+        ))
+        lines.append("")
+
+    return "\n".join(lines).rstrip() + "\n"
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("board_json", type=Path)
+    parser.add_argument("board_md", type=Path)
+    parser.add_argument("--check", action="store_true", help="fail if board_md is not up to date")
+    args = parser.parse_args()
+
+    board = load_board(args.board_json)
+    errors = validate_board(board)
+    if errors:
+        for error in errors:
+            print(f"ERROR: {error}", file=sys.stderr)
+        return 1
+    rendered = render(board)
+    if args.check:
+        existing = args.board_md.read_text() if args.board_md.exists() else ""
+        if existing != rendered:
+            print(f"ERROR: {args.board_md} is not up to date", file=sys.stderr)
+            return 1
+        print(f"PASS: {args.board_md} is up to date and roadmap coverage is complete")
+        return 0
+    args.board_md.parent.mkdir(parents=True, exist_ok=True)
+    args.board_md.write_text(rendered)
+    print(f"wrote {args.board_md}")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

.omx/cc2/validate_issue_parity_intake.py

@@ -0,0 +1,58 @@
+#!/usr/bin/env python3
+"""Validate the worker-2 CC2 issue/parity intake fragment."""
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[2]
+INTAKE = ROOT / ".omx" / "cc2" / "issue-parity-intake.json"
+REQUIRED_ISSUES = set(range(3028, 3039)) | {3007, 3006, 3020, 3005, 3003, 2997, 3023, 3004}
+ALLOWED_STATUS = {
+    "context",
+    "active",
+    "open",
+    "done_verify",
+    "stale_done",
+    "superseded",
+    "deferred_with_rationale",
+    "rejected_not_claw",
+}
+ALLOWED_BUCKETS = {"alpha_blocker", "beta_adoption", "ga_ecosystem", "post_2_0_research"}
+
+
+def require(condition: bool, message: str) -> None:
+    if not condition:
+        raise SystemExit(f"FAIL: {message}")
+
+
+def main() -> None:
+    data = json.loads(INTAKE.read_text())
+    issue_rows = data.get("issue_clusters", [])
+    parity_rows = data.get("parity_rows", [])
+
+    seen = {row.get("source_number") for row in issue_rows}
+    missing = sorted(REQUIRED_ISSUES - seen)
+    extra = sorted(seen - REQUIRED_ISSUES)
+    require(not missing, f"missing required issue rows: {missing}")
+    require(not extra, f"unexpected issue rows in scoped intake: {extra}")
+    require(len(issue_rows) == len(REQUIRED_ISSUES), "duplicate or missing issue row count")
+
+    ids = [row.get("id") for row in issue_rows + parity_rows]
+    require(len(ids) == len(set(ids)), "duplicate ids present")
+
+    for row in issue_rows + parity_rows:
+        row_id = row.get("id")
+        for field in ["source_anchor", "source_type", "release_bucket", "lifecycle_status", "dependencies", "verification_required"]:
+            require(row.get(field) not in (None, "", []), f"{row_id} missing {field}")
+        require(row["release_bucket"] in ALLOWED_BUCKETS, f"{row_id} invalid release_bucket {row['release_bucket']}")
+        require(row["lifecycle_status"] in ALLOWED_STATUS, f"{row_id} invalid lifecycle_status {row['lifecycle_status']}")
+        if row["lifecycle_status"] == "deferred_with_rationale":
+            require(row.get("deferral_rationale"), f"{row_id} deferred without rationale")
+
+    require(len(parity_rows) >= data["coverage"]["parity_rows_expected_minimum"], "not enough parity rows")
+    print(f"PASS issue/parity intake: {len(issue_rows)} issue rows, {len(parity_rows)} parity rows")
+
+
+if __name__ == "__main__":
+    main()

.port_sessions/b035f648d5b549aa836ea01f6727ec62.json

@@ -0,0 +1,8 @@
+{
+  "session_id": "b035f648d5b549aa836ea01f6727ec62",
+  "messages": [
+    "review MCP tool"
+  ],
+  "input_tokens": 3,
+  "output_tokens": 13
+}

.port_sessions/b234acb1eb8c486e80544ddc7e13e6d8.json

@@ -0,0 +1,9 @@
+{
+  "session_id": "b234acb1eb8c486e80544ddc7e13e6d8",
+  "messages": [
+    "review MCP tool",
+    "review MCP tool"
+  ],
+  "input_tokens": 6,
+  "output_tokens": 32
+}

.port_sessions/b67e062748f04e10ac5770df9285e4bd.json

@@ -0,0 +1,9 @@
+{
+  "session_id": "b67e062748f04e10ac5770df9285e4bd",
+  "messages": [
+    "review MCP tool",
+    "review MCP tool"
+  ],
+  "input_tokens": 6,
+  "output_tokens": 32
+}

.port_sessions/bb88fd20433840a8b19237e3f306c6e3.json

@@ -0,0 +1,9 @@
+{
+  "session_id": "bb88fd20433840a8b19237e3f306c6e3",
+  "messages": [
+    "review MCP tool",
+    "review MCP tool"
+  ],
+  "input_tokens": 6,
+  "output_tokens": 32
+}

CLAUDE.md

@@ -7,7 +7,7 @@ This file provides guidance to Claude Code (claude.ai/code) when working with co
 - Frameworks: none detected from the supported starter markers.
 
 ## Verification
-- Run Rust verification from `rust/`: `cargo fmt`, `cargo clippy --workspace --all-targets -- -D warnings`, `cargo test --workspace`
+- Run Rust verification from repo root: `scripts/fmt.sh --check`; for formatting use `scripts/fmt.sh`. Run Rust clippy/tests from `rust/`: `cargo clippy --workspace --all-targets -- -D warnings`, `cargo test --workspace`
 - `src/` and `tests/` are both present; update both surfaces together when behavior changes.
 
 ## Repository shape

README.md

@@ -192,6 +192,7 @@ cargo test --workspace
 - [`PARITY.md`](./PARITY.md) — parity status for the Rust port
 - [`rust/MOCK_PARITY_HARNESS.md`](./rust/MOCK_PARITY_HARNESS.md) — deterministic mock-service harness details
 - [`ROADMAP.md`](./ROADMAP.md) — active roadmap and open cleanup work
+- [`docs/g004-events-reports-contract.md`](./docs/g004-events-reports-contract.md) — Stream 2 lane event/report contract guidance for consumers
 - [`PHILOSOPHY.md`](./PHILOSOPHY.md) — why the project exists and how it is operated
 
 ## Ecosystem

docs/g002-security-verification-map.md

@@ -0,0 +1,185 @@
+# G002 alpha security map and verification plan
+
+Generated by `worker-4` for OMX team task 5 on 2026-05-14.
+
+## Scope and coordination
+
+- Active goal context: `G002-alpha-security` / Stream 6 day-one security and permissions gate.
+- Worker ownership: `worker-1` owns minimal implementation changes for workspace/path enforcement. `worker-4` owns this repository map, integration verification plan, changed-file/commit report, and exact verification evidence.
+- Boundary: this report does not mutate `.omx/ultragoal` and does not edit shared security/path tests.
+- Parallel probe status: three native subagents were spawned for repository map, test probe, and change-slice probe, but all failed before returning findings with `429 Too Many Requests`; local mapping below is based on direct repository inspection.
+
+## Current permission and path enforcement map
+
+### Runtime permission policy and enforcer
+
+- `rust/crates/runtime/src/permissions.rs`
+  - Owns the `PermissionMode` ordering and `PermissionPolicy` authorization contract.
+  - Existing tests cover read-only denial, workspace-write escalation, prompt approvals/denials, danger-full-access allowance, override recording, and required-mode reporting.
+  - Integration risk: any new dynamic file/path rule must preserve the existing `PermissionPolicy::authorize` semantics so prompt/override audit events remain stable.
+
+- `rust/crates/runtime/src/permission_enforcer.rs`
+  - `PermissionEnforcer::check`, `check_with_required_mode`, `check_file_write`, and `check_bash` convert policy outcomes into structured `EnforcementResult` payloads.
+  - `check_file_write` currently has the direct write gate for workspace-write mode.
+  - `is_within_workspace` is a string-prefix boundary check after simple relative-path joining; it does not canonicalize symlinks, `..`, Windows drive prefixes, or case variants.
+  - Existing tests cover read-only denial, workspace-write inside/outside paths, trailing slashes, root equality, bash read-only heuristics, prompt-mode denial payloads, and structured denied fields.
+
+### File tool path handling
+
+- `rust/crates/runtime/src/file_ops.rs`
+  - `read_file`, `write_file`, and `edit_file` normalize paths before filesystem operations but do not themselves require a workspace root.
+  - `read_file_in_workspace`, `write_file_in_workspace`, and `edit_file_in_workspace` exist as boundary-enforced wrappers.
+  - `validate_workspace_boundary` canonicalizes through the caller-provided resolved path and checks `starts_with(workspace_root)`.
+  - `is_symlink_escape` detects direct symlink escapes by comparing canonical target to canonical workspace root.
+  - Search tools (`glob_search`, `grep_search`) derive walk roots and prune heavy directories, but they are separate from the write enforcement path.
+  - Existing tests cover oversized/binary reads, workspace-boundary read rejection, symlink escape detection, glob brace expansion, ignored directories, and grep/glob behavior.
+
+### Bash command validation
+
+- `rust/crates/runtime/src/bash_validation.rs`
+  - `validate_command` runs mode validation, sed validation, destructive warning checks, then path validation.
+  - `validate_read_only` blocks write-like commands, state-modifying commands, write redirects, and mutating git subcommands in read-only mode.
+  - `validate_mode` warns when workspace-write commands appear to target hard-coded system paths.
+  - `validate_paths` warns for `../`, `~/`, and `$HOME` references; it is intentionally heuristic and does not resolve shell expansion or canonical targets.
+  - Existing tests cover read-only blockers, destructive warnings, sed in-place blocking, path traversal/home warnings, command classification, and full pipeline allow/block/warn outcomes.
+
+### Sandbox and diagnostics surfaces
+
+- `rust/crates/runtime/src/sandbox.rs`
+  - Owns container/sandbox status detection and workspace-only sandbox command construction.
+  - Relevant for day-one security because sandbox status must not overstate filesystem isolation.
+
+- `rust/crates/rusty-claude-cli/src/main.rs`
+  - Owns CLI permission-mode parsing, direct JSON/text diagnostic output, `/permissions`, `/status`, `/doctor`, and command dispatch paths.
+  - Existing CLI integration tests under `rust/crates/rusty-claude-cli/tests/` cover permission prompt scenarios and output-format contracts.
+
+- `rust/crates/rusty-claude-cli/tests/mock_parity_harness.rs`
+  - End-to-end harness includes `bash_permission_prompt_approved`, `bash_permission_prompt_denied`, read/write file allow/deny, and plugin workspace-write scenarios.
+
+## Existing G002-adjacent coverage
+
+- Unit-level permission coverage:
+  - `cargo test -p runtime permissions::tests`
+  - `cargo test -p runtime permission_enforcer::tests`
+  - `cargo test -p runtime bash_validation::tests`
+  - `cargo test -p runtime file_ops::tests`
+
+- CLI and integration coverage:
+  - `cargo test -p rusty-claude-cli --test mock_parity_harness`
+  - `cargo test -p rusty-claude-cli --test output_format_contract`
+  - `cargo test -p rusty-claude-cli --test cli_flags_and_config_defaults`
+
+- Board/report validation coverage:
+  - `python3 scripts/validate_cc2_board.py --board .omx/cc2/board.json`
+  - `python3 .omx/cc2/validate_issue_parity_intake.py .omx/cc2/issue-parity-intake.json`
+
+## Recommended safe work slices
+
+### Implementation lane (owned by worker-1 unless re-scoped)
+
+1. Replace string-prefix workspace boundary checks with canonical path comparison in the runtime enforcement path.
+   - Primary files: `rust/crates/runtime/src/permission_enforcer.rs`, possibly shared helper extraction from `rust/crates/runtime/src/file_ops.rs`.
+   - Regression cases: `../` traversal, symlink escape, root prefix collision (`/workspace` vs `/workspacex`), relative paths, trailing slash root equality.
+
+2. Ensure direct file tools call workspace-aware wrappers when active permission mode is `workspace-write`.
+   - Primary files: likely `rust/crates/runtime/src/mcp_tool_bridge.rs` and/or the runtime tool execution bridge that calls `file_ops`.
+   - Regression cases: direct read/write paths, missing parent creation, symlink parent escape, and error payload stability.
+
+3. Keep bash validation as a warning/classification layer unless a real shell-expansion resolver is introduced.
+   - Primary files: `rust/crates/runtime/src/bash_validation.rs`, `rust/crates/runtime/src/bash.rs`.
+   - Risk: heuristic parsing cannot faithfully resolve shell expansion, globs, aliases, or platform-specific path rules; avoid claiming hard enforcement unless execution sandbox or command resolver proves it.
+
+### Test lane (coordinate with worker-3/worker-1 before editing)
+
+1. Add unit regressions close to each enforcement function before changing behavior.
+   - `permission_enforcer.rs`: canonical path boundary and Windows-shaped path cases.
+   - `file_ops.rs`: write/edit workspace wrappers with symlink parent escapes and missing file parent canonicalization.
+   - `bash_validation.rs`: shell expansion/glob/path warnings remain warnings unless a resolver is introduced.
+
+2. Add at least one integration test proving the runtime bridge actually routes file tools through workspace enforcement, not only helper functions.
+   - Candidate: `rust/crates/rusty-claude-cli/tests/mock_parity_harness.rs` for direct write denial and no file created outside workspace.
+
+3. Preserve existing prompt/event visibility tests.
+   - Candidate surfaces: permission prompt scenarios in `mock_parity_harness.rs`, status/doctor JSON in `output_format_contract.rs`.
+
+### Docs/reporting lane (owned by worker-4)
+
+1. Keep this file as the integration handoff artifact for G002 mapping and verification.
+2. Report changed files and commits relative to `origin/main` so the leader can integrate worker branches deterministically.
+3. Include exact command evidence in the task lifecycle result.
+
+## Changed files relative to `origin/main` at map time
+
+The worktree currently contains these files added relative to `origin/main` before this task report:
+
+- `.omx/cc2/board.json`
+- `.omx/cc2/board.md`
+- `.omx/cc2/issue-parity-intake.json`
+- `.omx/cc2/issue-parity-intake.md`
+- `.omx/cc2/render_board_md.py`
+- `.omx/cc2/validate_issue_parity_intake.py`
+- `scripts/cc2_board.py`
+- `scripts/generate_cc2_board.py`
+- `scripts/validate_cc2_board.py`
+
+This task adds:
+
+- `docs/g002-security-verification-map.md`
+
+## Commits relative to `origin/main` at map time
+
+- `8311655` — `omx(team): auto-checkpoint worker-1 [1]`
+- `c6e2a7d` — `omx(team): merge worker-1`
+- `481585f` — `omx(team): auto-checkpoint worker-1 [1]`
+- `74bbf4b` — `omx(team): auto-checkpoint worker-4 [unknown]`
+- `5c77896` — `omx(team): auto-checkpoint worker-1 [1]`
+- `07dad88` — `Classify issue and parity intake for CC2 board integration`
+- `424825f` — `task: G001 human board and docs rendering`
+- `d15268e` — `Create a canonical CC2 board so every frozen ROADMAP heading is verifiably mapped`
+- `45b43b5` — `Make the CC2 board schema executable for G001`
+
+## Verification checklist for leader integration
+
+Run these from the repository root unless noted:
+
+1. Python board/schema validation:
+   - `python3 scripts/validate_cc2_board.py --board .omx/cc2/board.json`
+   - `python3 .omx/cc2/validate_issue_parity_intake.py .omx/cc2/issue-parity-intake.json`
+
+2. Rust formatting and lint/type checks:
+   - `scripts/fmt.sh --check`
+   - `(cd rust && cargo check --workspace)`
+   - `(cd rust && cargo clippy --workspace --all-targets -- -D warnings)`
+
+3. Targeted G002 security tests:
+   - `(cd rust && cargo test -p runtime permissions::tests permission_enforcer::tests bash_validation::tests file_ops::tests)`
+   - `(cd rust && cargo test -p rusty-claude-cli --test mock_parity_harness)`
+
+4. Full regression:
+   - `(cd rust && cargo test --workspace)`
+
+
+## Worker-4 verification evidence (2026-05-14)
+
+PASS:
+
+- `python3 scripts/validate_cc2_board.py --board .omx/cc2/board.json` → `PASS cc2 board validation`; 729 items; ROADMAP headings `124/124`; ROADMAP actions `542/542`.
+- `python3 .omx/cc2/validate_issue_parity_intake.py .omx/cc2/issue-parity-intake.json` → `PASS issue/parity intake: 19 issue rows, 9 parity rows`.
+- `scripts/fmt.sh --check` → no output and zero exit before Rust checks continued.
+- `(cd rust && cargo check --workspace)` → `Finished dev profile` successfully.
+- `(cd rust && cargo test -p runtime permissions::tests)` → 9 passed.
+- `(cd rust && cargo test -p runtime permission_enforcer::tests)` → 21 passed.
+- `(cd rust && cargo test -p runtime bash_validation::tests)` → 32 passed.
+- `(cd rust && cargo test -p runtime file_ops::tests)` → 14 passed.
+- `(cd rust && cargo test -p rusty-claude-cli --test mock_parity_harness)` → 1 passed.
+
+FAIL / integration blockers observed on this worktree:
+
+- `(cd rust && cargo clippy --workspace --all-targets -- -D warnings)` failed in existing runtime code, not this docs-only task:
+  - `rust/crates/runtime/src/compact.rs:215` / `:216`: `clippy::match_same_arms`.
+  - `rust/crates/runtime/src/policy_engine.rs:5`: `clippy::duration-suboptimal-units`.
+  - `rust/crates/runtime/src/sandbox.rs:295-302`: `clippy::map_unwrap_or`.
+- `(cd rust && cargo test --workspace)` failed after broad success in API/commands/plugins/runtime tests because `rusty-claude-cli` unit test `tests::session_lifecycle_prefers_running_process_over_idle_shell` asserted `RunningProcess` but observed `IdleShell`.
+- Rerun of the specific failing test confirmed deterministic failure: `(cd rust && cargo test -p rusty-claude-cli --bin claw tests::session_lifecycle_prefers_running_process_over_idle_shell -- --exact --nocapture)` → 0 passed, 1 failed with the same `IdleShell` vs `RunningProcess` assertion.
+
+Recommended owner for failures: not `worker-4` unless re-scoped. These failures are outside the docs/report artifact and touch shared runtime/CLI implementation files.

docs/g003-boot-session-verification-map.md

@@ -0,0 +1,96 @@
+# G003 boot/session/preflight verification map
+
+Generated by `worker-1` for OMX team task 2 on 2026-05-14.
+
+## Scope and coordination
+
+- Active goal context: `G003-boot-session` / Stream 1 reliable worker boot and session control.
+- Boundary: this artifact is an audit/integration map only. It does not mutate `.omx/ultragoal` and it does not change shared implementation or tests.
+- Current worker split from leader mailbox:
+  - `worker-1`: task 1 worker boot / prompt SLA plus this task 2 audit map.
+  - `worker-2`: default trusted roots / trust resolver.
+  - `worker-3`: startup-no-evidence classifier.
+  - `worker-4`: session control plus preflight/doctor JSON surfaces.
+- Native subagent probes were attempted for Task 2 (`test probe` and `debug/root-cause probe`) but both failed before returning findings with `429 Too Many Requests`; the map below is based on direct repository inspection.
+
+## Implementation surface map
+
+### Worker boot lifecycle and prompt SLA
+
+- `rust/crates/runtime/src/worker_boot.rs`
+  - Core state types: `WorkerStatus`, `WorkerFailureKind`, `WorkerEventKind`, `WorkerEventPayload`, `StartupFailureClassification`, `StartupEvidenceBundle`, `WorkerTaskReceipt`, and `WorkerReadySnapshot`.
+  - Control plane: `WorkerRegistry::{create,get,observe,resolve_trust,send_prompt,await_ready,restart,terminate,observe_completion,observe_startup_timeout}`.
+  - Lifecycle states currently covered in code: `spawning`, `trust_required`, `tool_permission_required`, `ready_for_prompt`, `running`, `finished`, and `failed`.
+  - Prompt delivery semantics currently use `Running` events and fields `prompt_in_flight`, `last_prompt`, `expected_receipt`, `replay_prompt`, and `prompt_delivery_attempts`.
+  - Startup-no-evidence surface: `observe_startup_timeout` builds `StartupEvidenceBundle` and classifies trust, tool permission, prompt acceptance timeout, prompt misdelivery, transport death, worker crash, or unknown.
+  - File observability surface: `emit_state_file` writes `.claw/worker-state.json` with status, readiness, trust state, prompt-in-flight flag, last event, and update age.
+
+- `rust/crates/tools/src/lib.rs`
+  - Tool APIs expose the worker control plane through `WorkerCreate`, `WorkerGet`, `WorkerObserve`, `WorkerResolveTrust`, `WorkerAwaitReady`, `WorkerSendPrompt`, `WorkerRestart`, `WorkerTerminate`, and `WorkerObserveCompletion`.
+  - `WorkerCreate` merges `ConfigLoader::trusted_roots()` with per-call `trusted_roots` before calling `WorkerRegistry::create`.
+  - Tool-level tests exercise worker create/observe/send/restart/terminate/completion and state-file transitions.
+
+### Trust resolver and default trusted roots
+
+- `rust/crates/runtime/src/trust_resolver.rs`
+  - `TrustConfig`, `TrustAllowlistEntry`, and `TrustResolver` model trust prompts, allowlist/denylist policy, auto-trust, manual approval, and emitted trust events.
+  - `path_matches_trusted_root` and internal `path_matches` canonicalize paths when possible.
+  - Hazard: prefix matching must avoid accidental sibling matches such as `/tmp/work` matching `/tmp/work-evil`; worker-2 owns any changes here.
+
+- `rust/crates/runtime/src/config.rs`
+  - `trustedRoots` is parsed by `parse_optional_trusted_roots` and exposed through `RuntimeConfig::trusted_roots()` / feature config accessors.
+  - Current default is empty when unset; any project default roots work belongs to worker-2.
+
+### Session control
+
+- `rust/crates/runtime/src/session_control.rs`
+  - `SessionStore` namespaces sessions by canonical workspace fingerprint.
+  - Key API: `from_cwd`, `from_data_dir`, `create_handle`, `resolve_reference`, `resolve_managed_path`, `list_sessions`, `latest_session`, `load_session`, and `fork_session`.
+  - Guardrail: `validate_loaded_session` rejects cross-workspace sessions and allows legacy sessions only when their path remains inside the current workspace.
+  - Worker-4 owns changes to this lane.
+
+### CLI doctor/status/preflight and bootstrap-adjacent surfaces
+
+- `rust/crates/commands/src/lib.rs`
+  - Slash command definitions include `/status`, `/sandbox`, and `/doctor`.
+  - JSON rendering for command surfaces exists through handler functions and tests in the same module.
+
+- `rust/crates/tools/src/lib.rs`
+  - Bash and PowerShell tool runners include `workspace_test_branch_preflight`, which returns structured output with `return_code_interpretation: preflight_blocked:branch_divergence` for broad workspace tests on stale branches.
+  - Tests around `bash_workspace_tests_are_blocked_when_branch_is_behind_main` and targeted-test skipping protect this preflight behavior.
+
+## Existing focused verification commands
+
+Run from `rust/` unless noted.
+
+- Worker boot runtime contract:
+  - `cargo test -p runtime worker_boot -- --nocapture`
+- Worker tool API contract:
+  - `cargo test -p tools worker_ -- --nocapture`
+- Session control contract:
+  - `cargo test -p runtime session_control -- --nocapture`
+- Trust resolver/config trusted roots:
+  - `cargo test -p runtime trust_resolver -- --nocapture`
+  - `cargo test -p runtime config::tests::parses_trusted_roots_from_settings config::tests::trusted_roots_default_is_empty_when_unset -- --nocapture`
+- Preflight/tool branch guardrails:
+  - `cargo test -p tools bash_workspace_tests_are_blocked_when_branch_is_behind_main bash_targeted_tests_skip_branch_preflight -- --nocapture`
+- Formatting/type/lint baseline:
+  - `../scripts/fmt.sh --check`
+  - `cargo check -p runtime -p tools -p commands`
+  - `cargo clippy -p runtime -p tools -p commands --all-targets --no-deps -- -D warnings`
+
+## Gaps and hazards for leader integration
+
+- Prompt SLA event naming is partially implicit: `send_prompt` emits `WorkerEventKind::Running`; it does not expose separate `prompt.sent`, `prompt.accepted`, `prompt.acceptance_delayed`, or `prompt.acceptance_timeout` event names. The current equivalent evidence is `prompt_in_flight`, `Running`, `observe_completion`, and startup-timeout classification.
+- `StartupFailureClassification::PromptAcceptanceTimeout` is covered in `worker_boot` tests; full terminal/transport integration should still be verified by the leader or worker-3 if a real pane watcher exists outside the in-memory registry.
+- Default trusted roots are parsed and merged into `WorkerCreate`, but unset config currently means no default roots. Worker-2 owns any change to default root selection.
+- Session control protects workspace fingerprints at load/fork time; worker-4 owns CLI/doctor/preflight JSON contract changes.
+- Full-workspace clippy currently has known unrelated runtime findings observed during task 1 verification; do not block this docs-only map on those unless leader re-scopes cleanup.
+
+## Recommended safe integration order
+
+1. Integrate worker boot / prompt SLA changes first and run `cargo test -p runtime worker_boot -- --nocapture` plus `cargo test -p tools worker_ -- --nocapture`.
+2. Integrate trust-root changes and rerun trust/config tests plus the worker create config merge test.
+3. Integrate startup-no-evidence classifier changes and rerun `cargo test -p runtime worker_boot -- --nocapture`.
+4. Integrate session control / preflight / doctor JSON changes and rerun session-control, commands JSON, and preflight tests.
+5. Run final formatting, targeted cargo check/clippy, then broader workspace tests with known full-workspace failures documented separately.

docs/g004-events-reports-contract.md

@@ -0,0 +1,67 @@
+# G004 event and report contract guidance
+
+Captured: 2026-05-14 during the Stream 2 `G004-events-reports` team run.
+
+Purpose: keep the user/developer-facing contract guidance for ROADMAP Phase 2 in one tracked source that points back to the code and roadmap anchors. This document is intentionally not the implementation map for task 5; it describes the interoperability contract consumers should rely on as the lane-event, report-schema, approval-token, and capability-negotiation lanes land.
+
+## Source-of-truth anchors
+
+| Contract family | Roadmap anchor | Current implementation / owner-facing anchor | Consumer guidance |
+| --- | --- | --- | --- |
+| Canonical lane events | `ROADMAP.md` Phase 2 §4, §4.5, §4.6, §4.7 | `rust/crates/runtime/src/lane_events.rs` (`LaneEventName`, `LaneEventStatus`, `LaneEventMetadata`, terminal reconciliation helpers) | Consume `event`, `status`, `emittedAt`, and `metadata` fields as the canonical state stream; do not infer lane state from terminal text when a structured event is present. |
+| Report schema v1 and projections | `ROADMAP.md` §4.25-§4.34 | Stream 2 report-schema lane / fixtures as they land | Treat a report as a versioned canonical payload plus derived projections. A projection may omit or transform fields only with explicit provenance: compatibility downgrade, redaction policy, truncation, or source absence. |
+| Policy-blocked handoff and approval-token chain | `ROADMAP.md` §4.37-§4.39 | Stream 2 approval-token lane as it lands | Treat policy blocks and owner approvals as typed artifacts, not prose. Execute an exception only when the approval token matches actor, policy, action, repo/branch/commit scope, expiry, and one-time-use state. |
+| Capability negotiation | `ROADMAP.md` §4.25, §4.26, §4.32, §4.34 | Report-schema/projection fixtures and consumer conformance cases as they land | Consumers must advertise supported schema versions, optional field families, projection views, redaction semantics, and downgrade handling before relying on reduced payloads. |
+
+## Lane event contract
+
+The lane-event stream is the first machine-trustworthy surface for Stream 2. Consumers should expect these invariants when reading `LaneEvent` payloads:
+
+- `event` is a typed event name, currently including the core lane lifecycle (`lane.started`, `lane.ready`, `lane.blocked`, `lane.red`, `lane.green`, `lane.finished`, `lane.failed`), branch health (`branch.stale_against_main`, `branch.workspace_mismatch`), reconciliation (`lane.reconciled`, `lane.superseded`, `lane.closed`), and ship provenance (`ship.prepared`, `ship.commits_selected`, `ship.merged`, `ship.pushed_main`).
+- `status` is the normalized state for the event; consumers should prefer it over freeform `detail` text for automation.
+- `metadata.seq`, `metadata.timestamp_ms`, and terminal fingerprints are the ordering/deduplication hooks. Consumers should use terminal reconciliation output rather than double-reporting contradictory terminal bursts.
+- `metadata.provenance`, `metadata.environment_label`, `metadata.emitter_identity`, and `metadata.confidence_level` tell consumers whether an event is live lane truth, test traffic, healthcheck/replay output, or transport-layer evidence.
+- `metadata.session_identity` and `metadata.ownership` bind a lane event to the session, workspace, workflow scope, owner, and watcher action. A watcher should not act on events whose ownership says `observe` or `ignore`.
+
+Minimal consumer rule: if a structured event exists, pane text is supporting evidence only. Pane scraping must not override a higher-confidence typed event with matching session/workflow ownership.
+
+## Report schema v1 contract
+
+A Stream 2 report should be treated as a canonical fact record with optional projections. Consumers should preserve these semantics even when they receive only a downgraded view:
+
+- Every report payload declares a schema version and a stable report identity/content hash for the full-fidelity canonical payload.
+- Assertions are labeled as `fact`, `hypothesis`, or another declared evidence class, with confidence and source references. Negative evidence is first-class: `not observed`, `checked and absent`, and `redacted` are distinct states.
+- Field deltas name the field, previous value/state, new value/state, attribution, and whether the delta came from source content, projection, downgrade, or redaction policy.
+- Projections carry lineage back to the canonical report id/content hash and name the projection view, capability set, schema version, redaction policy, and deterministic rendering inputs.
+- Redaction provenance is explicit. A missing field without a redaction/downgrade/source-absence reason is not enough evidence for an automated consumer to conclude the underlying fact is absent.
+
+Minimal consumer rule: store the canonical identity and projection metadata together. Do not compare two projections as state changes unless their canonical content hash or declared projection inputs differ.
+
+## Approval-token and policy-blocked contract
+
+Policy-blocked actions and owner-approved exceptions belong in the same structured event/report family:
+
+- A policy block names the typed reason, policy source, actor scope, blocked action, and safe fallback path.
+- An approval token names the approving actor, policy exception, action, repository/worktree/branch/commit scope, expiry, and allowed use count.
+- Token consumption records the exact action and scope that spent the token. Replays, scope expansion, expired tokens, and revoked tokens should surface typed policy errors.
+- Delegation traceability stays attached when another worker/lane executes the approved action; the executor must be able to prove which approval artifact authorized the exception.
+
+Minimal consumer rule: prose such as "approved" is not an executable approval. Require the structured token and verify that it is unconsumed and scoped to the exact action before proceeding.
+
+## Capability negotiation and conformance
+
+Mixed-version consumers are expected during Stream 2 rollout. Producers and consumers should negotiate instead of silently dropping fields:
+
+- Consumers advertise supported report schema versions, field families, projection views, redaction states, downgrade semantics, and fixture/conformance suite version.
+- Producers preserve one canonical full-fidelity report and emit downgraded projections only with `downgraded_for_compatibility` metadata.
+- Deterministic projection inputs include schema version, consumer capability set, projection policy version, redaction policy version, and canonical content hash.
+- Consumer conformance should distinguish syntax acceptance from semantic correctness, especially for `redacted` vs `missing`, stale vs current projections, negative evidence, and approval-token replay states.
+
+Minimal consumer rule: an older consumer may accept a downgraded projection, but it must surface the downgrade as a capability limitation rather than treating omitted fields as canonical absence.
+
+## Documentation maintenance rules
+
+- Keep ROADMAP Phase 2 as the product requirement source and this file as the contract-reading guide.
+- Keep Rust type names and event names aligned with `rust/crates/runtime/src/lane_events.rs`; update this document in the same change when public event names or metadata semantics change.
+- Keep report-schema examples/fixtures aligned with this guide once the schema lane lands; fixture updates should explain intentional schema or projection changes.
+- Do not mutate `.omx/ultragoal` from worker lanes. Leader-owned Ultragoal checkpointing consumes commits and verification evidence from task results.

docs/g004-events-reports-verification-map.md

@@ -0,0 +1,57 @@
+# G004 events/reports verification map
+
+Scope source: OMX team `g004-events-reports-u-e61d2271`, worker-1 tasks 1, 2, 4, 5. Workers must not mutate `.omx/ultragoal`; leader owns aggregate checkpoints.
+
+## Ownership boundaries
+
+- **Lane events / event identity / terminal reconciliation** — `rust/crates/runtime/src/lane_events.rs`, exported through `rust/crates/runtime/src/lib.rs`; tool-manifest consumers in `rust/crates/tools/src/lib.rs` write `LaneEvent` vectors.
+- **Report schema v1 / projection / redaction / capability negotiation** — `rust/crates/runtime/src/report_schema.rs`, exported through `rust/crates/runtime/src/lib.rs`; fixture note at `rust/crates/runtime/tests/fixtures/report_schema_v1/README.md`.
+- **Approval-token chain** — ROADMAP §§4.38-4.40; owned by worker-2 for this team split. Worker-1 did not edit it.
+- **Pinpoint closure batch** — runtime hygiene across compact/search-parser/policy/sandbox/integration-test surfaces: `rust/crates/runtime/src/compact.rs`, `rust/crates/runtime/src/file_ops.rs`, `rust/crates/runtime/src/policy_engine.rs`, `rust/crates/runtime/src/sandbox.rs`, `rust/crates/runtime/tests/integration_tests.rs`.
+- **Regression harness / docs alignment** — worker-3/worker-4 lanes per leader split. Coordinate before editing shared docs/tests.
+
+## Relevant symbols and files
+
+- `LaneEventName`, `LaneEventStatus`, `LaneEventMetadata`, `LaneEventBuilder`, `compute_event_fingerprint`, `dedupe_terminal_events`, `reconcile_terminal_events` in `runtime/src/lane_events.rs`.
+- `CanonicalReportV1`, `ReportClaim`, `NegativeEvidence`, `FieldDelta`, `ConsumerCapabilities`, `ReportProjectionV1`, `canonicalize_report`, `project_report`, `report_schema_v1_registry` in `runtime/src/report_schema.rs`.
+- `AgentOutput.lane_events`, `persist_agent_terminal_state`, `write_agent_manifest`, `maybe_commit_provenance` in `tools/src/lib.rs`.
+- Search/parser closure helpers: `summarize_messages` in `compact.rs`, `grep_search_impl` / `build_grep_content_output` in `file_ops.rs`.
+
+## Completed worker-1 commits
+
+- `f45f05e` / task 1 auto-checkpoint — terminal event fingerprints use stable SHA-256-derived canonical JSON, and production convenience terminal events attach/refresh fingerprints after payload changes.
+- `3989fc0` — report schema v1 contract, deterministic projection/redaction provenance, capability negotiation, and fixture note.
+- `7fff4c4` / task 4 auto-checkpoint — strict runtime clippy closure batch across compact/file_ops/policy/sandbox/integration tests.
+
+## Current verification evidence
+
+Run from `rust/` unless noted:
+
+- `cargo test -p runtime lane_events -- --nocapture` — PASS, 46 lane-event tests.
+- `cargo test -p runtime report_schema -- --nocapture` — PASS, 4 report-schema tests.
+- `cargo check -p runtime` — PASS.
+- `cargo clippy -p runtime --all-targets -- -D warnings` — PASS after task 4 closure batch.
+- `cargo test -p runtime -- --nocapture` — PASS, 531 unit tests, 12 integration tests, doc-tests pass.
+- `cargo test -p tools lane_event_schema_serializes_to_canonical_names -- --nocapture` — PASS, 1 targeted tools contract test.
+
+## Leader integration verification plan
+
+1. Inspect worker commits: `git log --oneline --decorate --max-count=8`.
+2. Re-run focused contracts:
+   - `cd rust && cargo test -p runtime lane_events -- --nocapture`
+   - `cd rust && cargo test -p runtime report_schema -- --nocapture`
+   - `cd rust && cargo test -p tools lane_event_schema_serializes_to_canonical_names -- --nocapture`
+3. Re-run runtime quality gate:
+   - `cd rust && cargo check -p runtime`
+   - `cd rust && cargo clippy -p runtime --all-targets -- -D warnings`
+   - `cd rust && cargo test -p runtime -- --nocapture`
+4. If merging with worker-2 approval-token work, additionally run the worker-2 focused approval-token tests and check for export conflicts in `runtime/src/lib.rs`.
+5. If merging with worker-3/4 docs or harness work, re-run their named regression harnesses plus `git diff --check`.
+
+## Integration hazards
+
+- `runtime/src/lib.rs` export blocks are shared; resolve conflicts by keeping both lane-event and report-schema exports sorted enough to remain readable.
+- `tools/src/lib.rs` serializes lane events into agent manifests; terminal fingerprint changes intentionally affect `metadata.event_fingerprint` for finished/failed/superseded/merged/closed events with payloads.
+- `report_schema.rs` currently defines the reusable contract and in-code deterministic fixtures; it does not yet wire report emission into CLI/status surfaces.
+- ROADMAP approval-token §§4.38-4.40 remain a separate lane; do not treat worker-1 report schema as an approval artifact.
+- Full workspace checks may include unrelated slow/provider-dependent tests; the verified local gate for this stream is runtime + targeted tools tests above.

docs/g005-branch-recovery-verification-map.md

@@ -0,0 +1,40 @@
+# G005 Branch Recovery Verification Map
+
+Scope: worker-1 follow-up map for G005 branch/test awareness and recovery. This file intentionally does not mutate leader-owned `.omx/ultragoal` state.
+
+## Covered ROADMAP / PRD pinpoints
+
+- `ROADMAP.md:912-921` — Phase 3 §7 stale-branch detection before broad verification: broad workspace test commands are preflighted before execution, stale/diverged branches emit `branch.stale_against_main`, and targeted tests bypass the broad-test gate.
+- `ROADMAP.md:922-933` — Phase 3 §8 recovery recipes: stale-branch recovery remains represented by the `stale_branch` recipe, with one automatic attempt before escalation.
+- `ROADMAP.md:935-949` — Phase 3 §8.5 recovery attempt ledger: `RecoveryContext` now exposes ledger entries with recipe id, attempt count, state, started/finished markers, last failure summary, and escalation reason.
+- `ROADMAP.md:951-970` — Phase 3 §9 green-ness / hung-test reporting: timed-out test commands now classify as `test.hung` with structured provenance instead of generic timeout.
+- `prd.json:37-44` — US-003 stale-branch detection before broad verification: verified through the `workspace_test_branch_preflight` broad-test block and targeted-test bypass tests.
+- `prd.json:50-57` — US-004 recovery recipes with ledger: verified through recovery ledger unit coverage and serialization-compatible recovery structs.
+
+## Implementation anchors
+
+- `rust/crates/runtime/src/stale_branch.rs` — existing branch freshness model and policy actions for fresh, stale, and diverged branches.
+- `rust/crates/tools/src/lib.rs` — `workspace_test_branch_preflight`, `branch_divergence_output`, Bash/PowerShell broad-test gating, and `test.hung` structured timeout provenance on tool-shell timeouts.
+- `rust/crates/runtime/src/recovery_recipes.rs` — recovery recipes plus `RecoveryLedgerEntry` / `RecoveryAttemptState` ledger surface.
+- `rust/crates/runtime/src/bash.rs` — runtime Bash timeout classification and structured provenance for hung test commands.
+- `rust/crates/runtime/src/lib.rs` — public exports for the recovery ledger types.
+
+## Verification evidence
+
+- `cargo test -p runtime` → PASS: 538 unit tests, 2 G004 conformance tests, 12 integration tests, and doctests passed.
+- `cargo test -p tools bash_tool_classifies_test_timeout_as_hung_with_provenance -- --nocapture` → PASS.
+- `cargo test -p tools bash_workspace_tests_are_blocked_when_branch_is_behind_main -- --nocapture` → PASS.
+- `cargo test -p tools bash_targeted_tests_skip_branch_preflight -- --nocapture` → PASS.
+- `cargo check -p runtime -p tools` → PASS.
+- `cargo clippy -p runtime --all-targets -- -D warnings` → PASS.
+- `cargo clippy -p tools --lib --no-deps -- -D warnings` → PASS.
+
+## Known unresolved / out-of-scope items
+
+- Full `cargo test -p tools` is still red on six permission-enforcer expectation tests unrelated to G005 branch freshness, recovery ledger, or hung-test classification. The failing tests assert old permission wording/read-only behavior and pre-existed this follow-up scope.
+- ROADMAP stale-base JSON/doctor/status pinpoints remain broader CLI diagnostic-surface work, especially `ROADMAP.md:2425-2489`, `ROADMAP.md:4346-4431`, and `ROADMAP.md:5061-5086`. They are related to branch freshness, but task 1 only required the broad-test freshness gate and narrow reporting surfaces.
+- No `.omx/ultragoal` files were changed; leader-owned Ultragoal checkpointing remains outside worker scope.
+
+## Delegation evidence
+
+Subagent spawn evidence: 1, Repository map probe `019e25d5-9be9-7193-8a33-f21450beb62c`; spawned before further serial task-2 mapping per contract, but errored with 429 Too Many Requests, so direct repo evidence was integrated instead.

docs/pr-issue-resolution-gate.md

@@ -0,0 +1,42 @@
+# Claw Code 2.0 PR and Issue Resolution Gate
+
+This gate was added to the Claw Code 2.0 Ultragoal after the explicit requirement:
+
+> all PRs should be merged and all issues should be resolved if resolvable and correct.
+
+## Scope
+
+Before the Claw Code 2.0 Ultragoal can be marked complete:
+
+1. Every open GitHub PR at the current final-gate snapshot must be triaged.
+2. PRs that are correct, compatible with Claw Code 2.0 direction, and pass required verification must be merged.
+3. PRs that are stale, incorrect, duplicative, unsafe, spam, or outside Claw Code scope must not be merged; each needs a recorded rationale.
+4. Every open GitHub issue at the current final-gate snapshot must be triaged.
+5. Issues that are resolvable and correct must be fixed or explicitly linked to a merged fix.
+6. Issues that are spam, duplicates, incorrect, unactionable, externally blocked, or not Claw Code work must be closed or labeled/commented with rationale when repository policy allows.
+7. The final completion audit must use a fresh GitHub snapshot, not only the planning snapshot.
+
+## Current live snapshot
+
+A live snapshot was captured locally during G002 execution:
+
+- PR snapshot: `.omx/research/github-live/open-prs.json`
+- Issue snapshot: `.omx/research/github-live/open-issues.json`
+- Captured on: 2026-05-14 during the active Ultragoal run.
+- Observed counts: 50 open PR records and 1000 open issue records from GitHub CLI list calls.
+
+These local `.omx/research/github-live/*` files are evidence inputs, not final proof. The final gate must refresh them and compare deltas.
+
+## Required final evidence
+
+The final report must include:
+
+- Fresh `gh pr list --state open` and `gh issue list --state open` snapshots.
+- A PR ledger with one row per PR: merge / reject / defer, reason, verification, commit/merge reference.
+- An issue ledger with one row per issue: fixed / duplicate / spam / invalid / deferred-with-rationale / externally-blocked, reason, and linked evidence.
+- Verification that no correct, mergeable PR remains unmerged without rationale.
+- Verification that no resolvable, correct issue remains open without a fix or rationale.
+
+## Non-goals
+
+This gate does not require merging unsafe, unverified, incompatible, spam, or incorrect contributions. It requires explicit evidence-backed triage and action for everything that is correct and resolvable.

docs/roadmap-pr-goals.md

@@ -0,0 +1,58 @@
+# Roadmap PR goal intake
+
+Captured: 2026-05-14 (Asia/Seoul) during the Claw Code 2.0 Ultragoal run.
+
+Purpose: make the user's follow-up requirement durable: all roadmap PRs should be merged when correct/resolvable, and unresolved roadmap deltas should become Ultragoal work rather than being lost. This file is a tracked companion to the leader-owned `.omx/ultragoal/goals.json` and `.omx/ultragoal/ledger.jsonl` artifacts.
+
+## Merge policy
+
+- Merge only PRs that are still relevant to Claw Code 2.0, are non-draft, target `main`, and are conflict-free after a fresh mergeability refresh.
+- Prefer squash merges with a Lore-style body when GitHub allows a direct PR merge.
+- If a PR is documentation-only but adds a real roadmap gap, merging it is acceptable once checks/conflicts are clean.
+- If a PR is stale, duplicated by already-landed work, or not product-aligned, do not force-merge; record the rationale and map any still-correct requirement into G011/G012.
+- After merging roadmap PRs, refresh generated board artifacts (`.omx/cc2/board.json`, `.omx/cc2/board.md`) so Stream 0 coverage stays current.
+
+## Open roadmap PRs with green historical checks
+
+These are first-pass merge candidates, pending fresh mergeability and conflict checks against current `main`.
+
+| PR | Title | Branch | Checks | Mergeable | URL |
+| --- | --- | --- | --- | --- | --- |
+| #2848 | docs(roadmap): add #333 — no in-session settings inspect command | `docs/roadmap-333-no-settings-inspect-command` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2848 |
+| #2846 | docs(roadmap): add #331 — export silently overwrites on repeated invocations | `docs/roadmap-331-export-filename-collision` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2846 |
+| #2869 | docs(roadmap): add #358 — history entries missing role field, no pagination | `docs/roadmap-348-history-entries-missing-role` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2869 |
+| #2850 | docs(roadmap): add #335 — session list omits created_at_ms field | `docs/roadmap-335-session-list-no-created-at` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2850 |
+| #2868 | docs(roadmap): add #356 — session list title always null; no rename command | `docs/roadmap-347-session-list-title-always-null` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2868 |
+| #2865 | docs(roadmap): add #362 — doctor auth false-positive: misses CLI session tokens | `docs/roadmap-345-doctor-auth-check-incomplete` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2865 |
+| #2864 | docs(roadmap): add #364 — /cost returns no cost_usd; identical to /stats | `docs/roadmap-344-cost-command-no-dollar-amount` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2864 |
+| #2867 | docs(roadmap): add #368 — export always appends .txt; response.file reflects mangled path | `docs/roadmap-346-export-forces-txt-extension` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2867 |
+| #2862 | docs(roadmap): add #342 — status json omits active session ID, workspace counters ambiguous | `docs/roadmap-342-v2` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2862 |
+| #2876 | docs(roadmap): add #354 — /cwd suggests itself in did-you-mean; self-referential loop | `docs/roadmap-354-cwd-self-referential-suggestion` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2876 |
+| #2872 | docs(roadmap): add #360 — /tokens, /stats, /cost identical output; no context-window or cost_usd | `docs/roadmap-349-tokens-stats-cost-identical` -> `main` | 4/4 checks successful | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2872 |
+
+## Open roadmap PRs needing local validation or CI refresh
+
+These have no check rollup in the live snapshot; validate locally or refresh CI before merging.
+
+| PR | Title | Branch | Checks | Mergeable | URL |
+| --- | --- | --- | --- | --- | --- |
+| #2858 | docs(roadmap): add #343 — session subcommand resume-safety inconsistently enforced | `docs/roadmap-340-session-resume-safe-inconsistent` -> `main` | no checks reported | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2858 |
+| #2839 | docs(roadmap): add #330 — resume mode stats/cost always zero | `docs/roadmap-324-resume-stats-zero` -> `main` | no checks reported | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2839 |
+| #2841 | docs(roadmap): add #332 — doctor json missing top-level status field | `docs/roadmap-325-doctor-no-status-field` -> `main` | no checks reported | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2841 |
+| #2844 | docs(roadmap): add #336 — session subcommand resume inconsistency and type/kind error mismatch | `docs/roadmap-329-session-subcommand-resume-inconsistency` -> `main` | no checks reported | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2844 |
+| #2842 | docs(roadmap): add #334 — version json omits build_date and uses short sha only | `docs/roadmap-328-version-json-incomplete` -> `main` | no checks reported | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2842 |
+
+## Product-fit review before merge
+
+These may be broader than the Claw Code 2.0 roadmap scope and need a product-fit decision before merge.
+
+| PR | Title | Branch | Checks | Mergeable | URL |
+| --- | --- | --- | --- | --- | --- |
+| #2824 | docs: personal assistant roadmap | `pr/docs-personal-assistant-roadmap` -> `main` | no checks reported | UNKNOWN | https://github.com/ultraworkers/claw-code/pull/2824 |
+
+## Ultragoal mapping
+
+- G003-G010: close implementation gaps that overlap a roadmap PR title if the requirement belongs to the active stream.
+- G011: reconcile ecosystem/ops/UX roadmap PRs and unresolved correct issues that do not fit earlier streams.
+- G012: final release gate must prove that every open roadmap PR was merged, closed as duplicate/obsolete, or converted into an explicit remaining goal with evidence.
+

progress.txt

@@ -74,6 +74,18 @@ US-007 COMPLETE (Phase 5 - Plugin/MCP lifecycle maturity)
 - DegradedMode behavior
 - Tests: 11 unit tests passing
 
+
+Iteration 2026-04-27 - ROADMAP #200 COMPLETED
+------------------------------------------------
+- Selected next actionable backlog item because no active task was in progress.
+- ROADMAP #200: Interactive MCP/tool permission prompts are invisible blockers.
+- Files: rust/crates/runtime/src/worker_boot.rs, rust/crates/runtime/src/recovery_recipes.rs, ROADMAP.md, progress.txt.
+- Added tool_permission_required worker status and event classification for interactive MCP/tool permission gates.
+- Added structured ToolPermissionPrompt payload with server/tool identity and prompt preview.
+- Startup evidence now records tool_permission_prompt_detected and classifies timeout evidence as tool_permission_required.
+- Readiness snapshots now mark tool-permission-gated workers as blocked, not ready/idle.
+- Tests: targeted tool_permission regressions, full runtime test/clippy/fmt pending in Ralph verification loop.
+
 VERIFICATION STATUS:
 ------------------
 - cargo build --workspace: PASSED
@@ -108,6 +120,29 @@ US-010 COMPLETED (Add model compatibility documentation)
 - Cross-referenced with existing code comments in openai_compat.rs
 - cargo clippy passes
 
+Iteration 3: 2026-04-16
+------------------------
+
+US-012 COMPLETED (Trust prompt resolver with allowlist auto-trust)
+- Files: rust/crates/runtime/src/trust_resolver.rs
+- Enhanced TrustConfig with pattern matching and serde support:
+  - TrustAllowlistEntry struct with pattern, worktree_pattern, description
+  - TrustResolution enum (AutoAllowlisted, ManualApproval)
+  - Enhanced TrustEvent variants with serde tags and metadata
+  - Glob pattern matching with * and ? wildcards
+  - Support for path prefix matching and worktree patterns
+- Updated TrustResolver with new resolve() signature:
+  - Added worktree parameter for worktree pattern matching
+  - Proper event emission with TrustResolution
+  - Manual approval detection from screen text
+- Added helper functions:
+  - extract_repo_name() - extracts repo name from path
+  - detect_manual_approval() - detects manual trust from screen text
+  - glob_matches() - recursive backtracking glob matcher
+- Tests: 25 new tests for pattern matching, serialization, and resolver behavior
+- All 483 runtime tests pass
+- cargo clippy passes with no warnings
+
 US-011 COMPLETED (Performance optimization: reduce API request serialization overhead)
 - Files:
   - rust/crates/api/Cargo.toml (added criterion dev-dependency and bench config)
@@ -131,3 +166,213 @@ US-011 COMPLETED (Performance optimization: reduce API request serialization ove
   - is_reasoning_model detection: ~26-42ns depending on model
 - All tests pass (119 unit tests + 29 integration tests)
 - cargo clippy passes
+
+VERIFICATION STATUS (Iteration 3):
+----------------------------------
+- cargo build --workspace: PASSED
+- cargo test --workspace: PASSED (891+ tests)
+- cargo clippy --workspace --all-targets -- -D warnings: PASSED
+- cargo fmt -- --check: PASSED
+
+All 12 stories from prd.json now have passes: true
+- US-001 through US-007: Pre-existing implementations
+- US-008: kimi-k2.5 model API compatibility fix
+- US-009: Unit tests for kimi model compatibility
+- US-010: Model compatibility documentation
+- US-011: Performance optimization with criterion benchmarks
+- US-012: Trust prompt resolver with allowlist auto-trust
+
+Iteration 4: 2026-04-16
+------------------------
+
+US-013 COMPLETED (Phase 2 - Session event ordering + terminal-state reconciliation)
+- Files: rust/crates/runtime/src/lane_events.rs
+- Added EventTerminality enum (Terminal, Advisory, Uncertainty)
+- Added classify_event_terminality() function for event classification
+- Added reconcile_terminal_events() function for deterministic event ordering:
+  - Sorts events by monotonic sequence number
+  - Deduplicates terminal events by fingerprint
+  - Detects transport death uncertainty (terminal + transport death)
+  - Handles out-of-order event bursts
+- Added events_materially_differ() for detecting meaningful differences
+- Added 8 comprehensive tests for reconciliation logic:
+  - reconcile_terminal_events_sorts_by_monotonic_sequence
+  - reconcile_terminal_events_deduplicates_same_fingerprint
+  - reconcile_terminal_events_detects_transport_death_uncertainty
+  - reconcile_terminal_events_handles_completed_idle_error_completed_noise
+  - reconcile_terminal_events_returns_none_for_empty_input
+  - reconcile_terminal_events_preserves_advisory_events
+  - events_materially_differ_detects_real_differences
+  - classify_event_terminality_correctly_classifies
+- Fixed test compilation issues with LaneEventBuilder API
+
+VERIFICATION STATUS (Iteration 4):
+----------------------------------
+- cargo build --workspace: PASSED
+- cargo test --workspace: PASSED (891+ tests)
+- cargo clippy --workspace --all-targets -- -D warnings: PASSED
+- cargo fmt -- --check: PASSED
+
+US-013 marked passes: true in prd.json
+
+US-014 COMPLETED (Phase 2 - Event provenance / environment labeling)
+- Files: rust/crates/runtime/src/lane_events.rs
+- Added ConfidenceLevel enum (High, Medium, Low, Unknown)
+- Added fields to LaneEventMetadata:
+  - environment_label: Option<String> - environment/channel (production, staging, dev)
+  - emitter_identity: Option<String> - emitter (clawd, plugin-name, operator-id)
+  - confidence_level: Option<ConfidenceLevel> - trust level for automation
+- Added builder methods: with_environment(), with_emitter(), with_confidence()
+- Added filtering functions:
+  - filter_by_provenance() - select events by source
+  - filter_by_environment() - select events by environment label
+  - filter_by_confidence() - select events above confidence threshold
+  - is_test_event() - check if synthetic source (test, healthcheck, replay)
+  - is_live_lane_event() - check if production event
+- Added 7 comprehensive tests for US-014:
+  - confidence_level_round_trips_through_serialization
+  - filter_by_provenance_selects_only_matching_events
+  - filter_by_environment_selects_only_matching_environment
+  - filter_by_confidence_selects_events_above_threshold
+  - is_test_event_detects_synthetic_sources
+  - is_live_lane_event_detects_production_events
+  - lane_event_metadata_includes_us014_fields
+
+US-016 COMPLETED (Phase 2 - Duplicate terminal-event suppression)
+- Files: rust/crates/runtime/src/lane_events.rs
+- Event fingerprinting already implemented via compute_event_fingerprint()
+- Fingerprint attached via LaneEventMetadata.event_fingerprint
+- Deduplication via dedupe_terminal_events() - returns first occurrence of each fingerprint
+- Raw event history preserved separately from deduplicated actionable events
+- Material difference detection via events_materially_differ():
+  - Different event type (Finished vs Failed) is material
+  - Different status is material
+  - Different failure class is material
+  - Different data payload is material
+- Reconcile function surfaces latest terminal event when materially different
+- Added 5 comprehensive tests for US-016:
+  - canonical_terminal_event_fingerprint_attached_to_metadata
+  - dedupe_terminal_events_suppresses_repeated_fingerprints
+  - dedupe_preserves_raw_event_history_separately
+  - events_materially_differ_detects_payload_differences
+  - reconcile_terminal_events_surfaces_latest_when_different
+
+US-017 COMPLETED (Phase 2 - Lane ownership / scope binding)
+- Files: rust/crates/runtime/src/lane_events.rs
+- LaneOwnership struct already existed with:
+  - owner: String - owner/assignee identity
+  - workflow_scope: String - workflow scope (claw-code-dogfood, etc.)
+  - watcher_action: WatcherAction - Act, Observe, Ignore
+- Ownership preserved through lifecycle via with_ownership() builder method
+- All lifecycle events (Started -> Ready -> Finished) preserve ownership
+- Added 3 comprehensive tests for US-017:
+  - lane_ownership_attached_to_metadata
+  - lane_ownership_preserved_through_lifecycle_events
+  - lane_ownership_watcher_action_variants
+
+US-015 COMPLETED (Phase 2 - Session identity completeness at creation time)
+- Files: rust/crates/runtime/src/lane_events.rs
+- SessionIdentity struct already existed with:
+  - title: String - stable title for the session
+  - workspace: String - workspace/worktree path
+  - purpose: String - lane/session purpose
+  - placeholder_reason: Option<String> - reason for placeholder values
+- Added reconcile_enriched() method for updating session identity:
+  - Updates title/workspace/purpose with newly available data
+  - Clears placeholder_reason when real values are provided
+  - Preserves existing values for fields not being updated
+  - Allows incremental enrichment without ambiguity
+- Added 2 comprehensive tests:
+  - session_identity_reconcile_enriched_updates_fields
+  - session_identity_reconcile_preserves_placeholder_if_no_new_data
+
+US-018 COMPLETED (Phase 2 - Nudge acknowledgment / dedupe contract)
+- Files: rust/crates/runtime/src/lane_events.rs
+- Added NudgeTracking struct:
+  - nudge_id: String - unique nudge identifier
+  - delivered_at: String - timestamp of delivery
+  - acknowledged: bool - whether acknowledged
+  - acknowledged_at: Option<String> - when acknowledged
+  - is_retry: bool - whether this is a retry
+  - original_nudge_id: Option<String> - original ID if retry
+- Added NudgeClassification enum (New, Retry, StaleDuplicate)
+- Added classify_nudge() function for deduplication logic
+- Added 6 comprehensive tests for US-018
+
+US-019 COMPLETED (Phase 2 - Stable roadmap-id assignment)
+- Files: rust/crates/runtime/src/lane_events.rs
+- Added RoadmapId struct:
+  - id: String - canonical unique identifier
+  - filed_at: String - timestamp when filed
+  - is_new_filing: bool - new vs update
+  - supersedes: Option<String> - lineage for supersedes
+- Added builder methods: new_filing(), update(), supersedes()
+- Added 3 comprehensive tests for US-019
+
+US-020 COMPLETED (Phase 2 - Roadmap item lifecycle state contract)
+- Files: rust/crates/runtime/src/lane_events.rs
+- Added RoadmapLifecycleState enum (Filed, Acknowledged, InProgress, Blocked, Done, Superseded)
+- Added RoadmapLifecycle struct:
+  - state: RoadmapLifecycleState - current state
+  - state_changed_at: String - last transition timestamp
+  - filed_at: String - original filing timestamp
+  - lineage: Vec<String> - supersession chain
+- Added methods: new_filed(), transition(), superseded_by(), is_terminal(), is_active()
+- Added 5 comprehensive tests for US-020
+
+VERIFICATION STATUS (Iteration 7):
+----------------------------------
+- cargo build --workspace: PASSED
+- cargo test --workspace: PASSED (891+ tests)
+- cargo clippy --workspace --all-targets -- -D warnings: PASSED
+- cargo fmt -- --check: PASSED
+
+US-013 through US-015 and US-018 through US-020 now marked passes: true
+
+FINAL VERIFICATION (All 20 Stories Complete):
+------------------------------------------------
+- cargo build --workspace: PASSED
+- cargo test --workspace: PASSED (119+ API tests, 39 runtime tests, 12 integration tests)
+- cargo clippy --workspace --all-targets -- -D warnings: PASSED
+- cargo fmt -- --check: PASSED
+
+ALL 20 STORIES FROM PRD COMPLETE:
+- US-001 through US-012: Pre-existing implementations (verified working)
+- US-013: Session event ordering + terminal-state reconciliation
+- US-014: Event provenance / environment labeling
+- US-015: Session identity completeness at creation time
+- US-016: Duplicate terminal-event suppression
+- US-017: Lane ownership / scope binding
+- US-018: Nudge acknowledgment / dedupe contract
+- US-019: Stable roadmap-id assignment
+- US-020: Roadmap item lifecycle state contract
+
+Iteration 8: 2026-04-16
+------------------------
+
+US-021 COMPLETED (Request body size pre-flight check - from dogfood findings)
+- Files:
+  - rust/crates/api/src/error.rs (new error variant)
+  - rust/crates/api/src/providers/openai_compat.rs
+- Added RequestBodySizeExceeded error variant with actionable message
+- Added max_request_body_bytes to OpenAiCompatConfig:
+  - DashScope: 6MB (6_291_456 bytes) - from dogfood with kimi-k2.5
+  - OpenAI: 100MB (104_857_600 bytes)
+  - xAI: 50MB (52_428_800 bytes)
+- Added estimate_request_body_size() for pre-flight checks
+- Added check_request_body_size() for validation
+- Pre-flight check integrated in send_raw_request()
+- Tests: 5 new tests for size estimation and limit checking
+
+PROJECT STATUS: COMPLETE (21/21 stories)
+
+Iteration 2026-04-29 - ROADMAP #96 COMPLETED
+------------------------------------------------
+- Pulled origin/main: already up to date.
+- Selected ROADMAP #96 as a small repo-local Immediate Backlog item: the `claw --help` Resume-safe command summary leaked slash-command stubs despite the main Interactive command listing filtering them.
+- Files: rust/crates/rusty-claude-cli/src/main.rs, ROADMAP.md, progress.txt.
+- Changed help rendering to filter `resume_supported_slash_commands()` through `STUB_COMMANDS` before building the Resume-safe one-liner.
+- Added `stub_commands_absent_from_resume_safe_help` regression coverage so future stub additions cannot leak into the Resume-safe summary.
+- Targeted verification: `cargo test -p rusty-claude-cli stub_commands_absent_from_resume_safe_help -- --nocapture` passed; `cargo test -p rusty-claude-cli parses_direct_cli_actions -- --nocapture` passed.
+- Format/check verification: `cargo fmt --all --check`, `git diff --check`, and `cargo check -p rusty-claude-cli` passed.
+- Broader clippy note: `cargo clippy -p rusty-claude-cli --all-targets -- -D warnings` is blocked by pre-existing `clippy::unnecessary_wraps` failures in `rust/crates/commands/src/lib.rs` (`render_mcp_report_for`, `render_mcp_report_json_for`), outside this diff.

rust/CLAUDE.md

@@ -7,7 +7,8 @@ This file provides guidance to Claw Code (clawcode.dev) when working with code i
 - Frameworks: none detected from the supported starter markers.
 
 ## Verification
-- Run Rust verification from the repo root: `cargo fmt`, `cargo clippy --workspace --all-targets -- -D warnings`, `cargo test --workspace`
+- From the repository root, run Rust formatting with `scripts/fmt.sh` (or `scripts/fmt.sh --check` for CI-style checks). From this `rust/` directory, the equivalent command is `../scripts/fmt.sh`. Root-level `cargo fmt --manifest-path rust/Cargo.toml` is not the supported formatting command.
+- From this `rust/` directory, run Rust verification with `cargo clippy --workspace --all-targets -- -D warnings` and `cargo test --workspace`.
 
 ## Working agreement
 - Prefer small, reviewable changes and keep generated bootstrap files aligned with actual repo workflows.

rust/crates/api/src/error.rs

@@ -14,6 +14,11 @@ const CONTEXT_WINDOW_ERROR_MARKERS: &[&str] = &[
     "too many tokens",
     "prompt is too long",
     "input is too long",
+    "input tokens exceed",
+    "configured limit",
+    "messages resulted in",
+    "completion tokens",
+    "prompt tokens",
     "request is too large",
 ];
 
@@ -542,6 +547,26 @@ mod tests {
         assert_eq!(error.request_id(), Some("req_ctx_123"));
     }
 
+    #[test]
+    fn classifies_openai_configured_limit_errors_as_context_window_failures() {
+        let error = ApiError::Api {
+            status: reqwest::StatusCode::BAD_REQUEST,
+            error_type: Some("invalid_request_error".to_string()),
+            message: Some(
+                "Input tokens exceed the configured limit of 922000 tokens. Your messages resulted in 1860900 tokens. Please reduce the length of the messages."
+                    .to_string(),
+            ),
+            request_id: Some("req_ctx_openai_123".to_string()),
+            body: String::new(),
+            retryable: false,
+            suggested_action: None,
+        };
+
+        assert!(error.is_context_window_failure());
+        assert_eq!(error.safe_failure_class(), "context_window");
+        assert_eq!(error.request_id(), Some("req_ctx_openai_123"));
+    }
+
     #[test]
     fn missing_credentials_without_hint_renders_the_canonical_message() {
         // given

rust/crates/api/src/lib.rs

@@ -21,11 +21,12 @@ pub use prompt_cache::{
 pub use providers::anthropic::{AnthropicClient, AnthropicClient as ApiClient, AuthSource};
 pub use providers::openai_compat::{
     build_chat_completion_request, flatten_tool_result_content, is_reasoning_model,
-    model_rejects_is_error_field, translate_message, OpenAiCompatClient, OpenAiCompatConfig,
+    model_rejects_is_error_field, model_requires_reasoning_content_in_history, translate_message,
+    OpenAiCompatClient, OpenAiCompatConfig,
 };
 pub use providers::{
     detect_provider_kind, max_tokens_for_model, max_tokens_for_model_with_override,
-    resolve_model_alias, ProviderKind,
+    model_family_identity_for, model_family_identity_for_kind, resolve_model_alias, ProviderKind,
 };
 pub use sse::{parse_frame, SseParser};
 pub use types::{

rust/crates/api/src/providers/mod.rs

@@ -250,19 +250,31 @@ pub fn detect_provider_kind(model: &str) -> ProviderKind {
     ProviderKind::Anthropic
 }
 
+#[must_use]
+pub const fn model_family_identity_for_kind(kind: ProviderKind) -> runtime::ModelFamilyIdentity {
+    match kind {
+        ProviderKind::Anthropic => runtime::ModelFamilyIdentity::Claude,
+        ProviderKind::Xai | ProviderKind::OpenAi => runtime::ModelFamilyIdentity::Generic,
+    }
+}
+
+#[must_use]
+pub fn model_family_identity_for(model: &str) -> runtime::ModelFamilyIdentity {
+    model_family_identity_for_kind(detect_provider_kind(model))
+}
+
 #[must_use]
 pub fn max_tokens_for_model(model: &str) -> u32 {
-    model_token_limit(model).map_or_else(
+    let canonical = resolve_model_alias(model);
-        || {
+    let heuristic = if canonical.contains("opus") {
-            let canonical = resolve_model_alias(model);
+        32_000
-            if canonical.contains("opus") {
+    } else {
-                32_000
+        64_000
-            } else {
+    };
-                64_000
+
-            }
+    model_token_limit(model)
-        },
+        .map(|limit| heuristic.min(limit.max_output_tokens))
-        |limit| limit.max_output_tokens,
+        .unwrap_or(heuristic)
-    )
 }
 
 /// Returns the effective max output tokens for a model, preferring a plugin
@@ -276,7 +288,8 @@ pub fn max_tokens_for_model_with_override(model: &str, plugin_override: Option<u
 #[must_use]
 pub fn model_token_limit(model: &str) -> Option<ModelTokenLimit> {
     let canonical = resolve_model_alias(model);
-    match canonical.as_str() {
+    let base_model = canonical.rsplit('/').next().unwrap_or(canonical.as_str());
+    match base_model {
         "claude-opus-4-6" => Some(ModelTokenLimit {
             max_output_tokens: 32_000,
             context_window_tokens: 200_000,
@@ -289,6 +302,20 @@ pub fn model_token_limit(model: &str) -> Option<ModelTokenLimit> {
             max_output_tokens: 64_000,
             context_window_tokens: 131_072,
         }),
+        // GPT-4.1 family via the OpenAI API.
+        "gpt-4.1" | "gpt-4.1-mini" | "gpt-4.1-nano" => Some(ModelTokenLimit {
+            max_output_tokens: 32_768,
+            context_window_tokens: 1_047_576,
+        }),
+        // GPT-5.4 family via the OpenAI API.
+        "gpt-5.4" => Some(ModelTokenLimit {
+            max_output_tokens: 128_000,
+            context_window_tokens: 1_000_000,
+        }),
+        "gpt-5.4-mini" | "gpt-5.4-nano" => Some(ModelTokenLimit {
+            max_output_tokens: 128_000,
+            context_window_tokens: 400_000,
+        }),
         // Kimi models via DashScope (Moonshot AI)
         // Source: https://platform.moonshot.cn/docs/intro
         "kimi-k2.5" | "kimi-k1.5" => Some(ModelTokenLimit {
@@ -470,8 +497,8 @@ mod tests {
     use super::{
         anthropic_missing_credentials, anthropic_missing_credentials_hint, detect_provider_kind,
         load_dotenv_file, max_tokens_for_model, max_tokens_for_model_with_override,
-        model_token_limit, parse_dotenv, preflight_message_request, resolve_model_alias,
+        model_family_identity_for, model_family_identity_for_kind, model_token_limit, parse_dotenv,
-        ProviderKind,
+        preflight_message_request, resolve_model_alias, ProviderKind,
     };
 
     /// Serializes every test in this module that mutates process-wide
@@ -530,6 +557,42 @@ mod tests {
         );
     }
 
+    #[test]
+    fn maps_provider_kind_to_model_family_identity() {
+        // given: each supported provider kind
+        let anthropic = ProviderKind::Anthropic;
+        let openai = ProviderKind::OpenAi;
+        let xai = ProviderKind::Xai;
+
+        // when: converting provider kinds to prompt model family identities
+        let anthropic_identity = model_family_identity_for_kind(anthropic);
+        let openai_identity = model_family_identity_for_kind(openai);
+        let xai_identity = model_family_identity_for_kind(xai);
+
+        // then: Anthropic stays Claude and OpenAI-compatible providers are generic
+        assert_eq!(anthropic_identity, runtime::ModelFamilyIdentity::Claude);
+        assert_eq!(openai_identity, runtime::ModelFamilyIdentity::Generic);
+        assert_eq!(xai_identity, runtime::ModelFamilyIdentity::Generic);
+    }
+
+    #[test]
+    fn maps_model_name_to_model_family_identity() {
+        // given: Anthropic, OpenAI-compatible, and xAI model names
+        let claude_model = "claude-opus-4-6";
+        let openai_model = "openai/gpt-4.1-mini";
+        let xai_model = "grok-3";
+
+        // when: detecting prompt model family identities from model names
+        let claude_identity = model_family_identity_for(claude_model);
+        let openai_identity = model_family_identity_for(openai_model);
+        let xai_identity = model_family_identity_for(xai_model);
+
+        // then: Anthropic stays Claude and OpenAI-compatible providers are generic
+        assert_eq!(claude_identity, runtime::ModelFamilyIdentity::Claude);
+        assert_eq!(openai_identity, runtime::ModelFamilyIdentity::Generic);
+        assert_eq!(xai_identity, runtime::ModelFamilyIdentity::Generic);
+    }
+
     #[test]
     fn openai_namespaced_model_routes_to_openai_not_anthropic() {
         // Regression: "openai/gpt-4.1-mini" was misrouted to Anthropic when
@@ -614,6 +677,15 @@ mod tests {
     fn keeps_existing_max_token_heuristic() {
         assert_eq!(max_tokens_for_model("opus"), 32_000);
         assert_eq!(max_tokens_for_model("grok-3"), 64_000);
+        assert_eq!(max_tokens_for_model("gpt-5.4"), 64_000);
+    }
+
+    #[test]
+    fn caps_default_max_tokens_to_openai_model_limits() {
+        assert_eq!(max_tokens_for_model("gpt-4.1-mini"), 32_768);
+        assert_eq!(max_tokens_for_model("openai/gpt-4.1-mini"), 32_768);
+        assert_eq!(max_tokens_for_model("gpt-5.4"), 64_000);
+        assert_eq!(max_tokens_for_model("openai/gpt-5.4"), 64_000);
     }
 
     #[test]
@@ -680,6 +752,18 @@ mod tests {
                 .context_window_tokens,
             131_072
         );
+        assert_eq!(
+            model_token_limit("openai/gpt-4.1-mini")
+                .expect("openai/gpt-4.1-mini should be registered")
+                .context_window_tokens,
+            1_047_576
+        );
+        assert_eq!(
+            model_token_limit("gpt-5.4")
+                .expect("gpt-5.4 should be registered")
+                .context_window_tokens,
+            1_000_000
+        );
     }
 
     #[test]
@@ -728,6 +812,42 @@ mod tests {
         }
     }
 
+    #[test]
+    fn preflight_blocks_oversized_requests_for_gpt_5_4() {
+        let request = MessageRequest {
+            model: "gpt-5.4".to_string(),
+            max_tokens: 64_000,
+            messages: vec![InputMessage {
+                role: "user".to_string(),
+                content: vec![InputContentBlock::Text {
+                    text: "x".repeat(3_900_000),
+                }],
+            }],
+            system: Some("Keep the answer short.".to_string()),
+            tools: None,
+            tool_choice: None,
+            stream: true,
+            ..Default::default()
+        };
+
+        let error = preflight_message_request(&request)
+            .expect_err("oversized gpt-5.4 request should be rejected before the provider call");
+
+        match error {
+            ApiError::ContextWindowExceeded {
+                model,
+                requested_output_tokens,
+                context_window_tokens,
+                ..
+            } => {
+                assert_eq!(model, "gpt-5.4");
+                assert_eq!(requested_output_tokens, 64_000);
+                assert_eq!(context_window_tokens, 1_000_000);
+            }
+            other => panic!("expected context-window preflight failure, got {other:?}"),
+        }
+    }
+
     #[test]
     fn preflight_skips_unknown_models() {
         let request = MessageRequest {
@@ -753,14 +873,14 @@ mod tests {
     #[test]
     fn returns_context_window_metadata_for_kimi_models() {
         // kimi-k2.5
-        let k25_limit = model_token_limit("kimi-k2.5")
+        let k25_limit =
-            .expect("kimi-k2.5 should have token limit metadata");
+            model_token_limit("kimi-k2.5").expect("kimi-k2.5 should have token limit metadata");
         assert_eq!(k25_limit.max_output_tokens, 16_384);
         assert_eq!(k25_limit.context_window_tokens, 256_000);
 
         // kimi-k1.5
-        let k15_limit = model_token_limit("kimi-k1.5")
+        let k15_limit =
-            .expect("kimi-k1.5 should have token limit metadata");
+            model_token_limit("kimi-k1.5").expect("kimi-k1.5 should have token limit metadata");
         assert_eq!(k15_limit.max_output_tokens, 16_384);
         assert_eq!(k15_limit.context_window_tokens, 256_000);
     }
@@ -768,11 +888,13 @@ mod tests {
     #[test]
     fn kimi_alias_resolves_to_kimi_k25_token_limits() {
         // The "kimi" alias resolves to "kimi-k2.5" via resolve_model_alias()
-        let alias_limit = model_token_limit("kimi")
+        let alias_limit =
-            .expect("kimi alias should resolve to kimi-k2.5 limits");
+            model_token_limit("kimi").expect("kimi alias should resolve to kimi-k2.5 limits");
-        let direct_limit = model_token_limit("kimi-k2.5")
+        let direct_limit = model_token_limit("kimi-k2.5").expect("kimi-k2.5 should have limits");
-            .expect("kimi-k2.5 should have limits");
+        assert_eq!(
-        assert_eq!(alias_limit.max_output_tokens, direct_limit.max_output_tokens);
+            alias_limit.max_output_tokens,
+            direct_limit.max_output_tokens
+        );
         assert_eq!(
             alias_limit.context_window_tokens,
             direct_limit.context_window_tokens

rust/crates/api/src/providers/openai_compat.rs

@@ -443,6 +443,8 @@ struct StreamState {
     stop_reason: Option<String>,
     usage: Option<Usage>,
     tool_calls: BTreeMap<u32, ToolCallState>,
+    thinking_started: bool,
+    thinking_finished: bool,
 }
 
 impl StreamState {
@@ -456,6 +458,8 @@ impl StreamState {
             stop_reason: None,
             usage: None,
             tool_calls: BTreeMap::new(),
+            thinking_started: false,
+            thinking_finished: false,
         }
     }
 
@@ -493,35 +497,61 @@ impl StreamState {
         }
 
         for choice in chunk.choices {
+            if let Some(reasoning) = choice
+                .delta
+                .reasoning_content
+                .filter(|value| !value.is_empty())
+            {
+                if !self.thinking_started {
+                    self.thinking_started = true;
+                    events.push(StreamEvent::ContentBlockStart(ContentBlockStartEvent {
+                        index: 0,
+                        content_block: OutputContentBlock::Thinking {
+                            thinking: String::new(),
+                            signature: None,
+                        },
+                    }));
+                }
+                events.push(StreamEvent::ContentBlockDelta(ContentBlockDeltaEvent {
+                    index: 0,
+                    delta: ContentBlockDelta::ThinkingDelta {
+                        thinking: reasoning,
+                    },
+                }));
+            }
+
             if let Some(content) = choice.delta.content.filter(|value| !value.is_empty()) {
+                self.close_thinking(&mut events);
                 if !self.text_started {
                     self.text_started = true;
                     events.push(StreamEvent::ContentBlockStart(ContentBlockStartEvent {
-                        index: 0,
+                        index: self.text_block_index(),
                         content_block: OutputContentBlock::Text {
                             text: String::new(),
                         },
                     }));
                 }
                 events.push(StreamEvent::ContentBlockDelta(ContentBlockDeltaEvent {
-                    index: 0,
+                    index: self.text_block_index(),
                     delta: ContentBlockDelta::TextDelta { text: content },
                 }));
             }
 
             for tool_call in choice.delta.tool_calls {
+                self.close_thinking(&mut events);
+                let tool_index_offset = self.tool_index_offset();
                 let state = self.tool_calls.entry(tool_call.index).or_default();
                 state.apply(tool_call);
-                let block_index = state.block_index();
+                let block_index = state.block_index(tool_index_offset);
                 if !state.started {
-                    if let Some(start_event) = state.start_event()? {
+                    if let Some(start_event) = state.start_event(tool_index_offset)? {
                         state.started = true;
                         events.push(StreamEvent::ContentBlockStart(start_event));
                     } else {
                         continue;
                     }
                 }
-                if let Some(delta_event) = state.delta_event() {
+                if let Some(delta_event) = state.delta_event(tool_index_offset) {
                     events.push(StreamEvent::ContentBlockDelta(delta_event));
                 }
                 if choice.finish_reason.as_deref() == Some("tool_calls") && !state.stopped {
@@ -535,11 +565,12 @@ impl StreamState {
             if let Some(finish_reason) = choice.finish_reason {
                 self.stop_reason = Some(normalize_finish_reason(&finish_reason));
                 if finish_reason == "tool_calls" {
+                    let tool_index_offset = self.tool_index_offset();
                     for state in self.tool_calls.values_mut() {
                         if state.started && !state.stopped {
                             state.stopped = true;
                             events.push(StreamEvent::ContentBlockStop(ContentBlockStopEvent {
-                                index: state.block_index(),
+                                index: state.block_index(tool_index_offset),
                             }));
                         }
                     }
@@ -557,19 +588,21 @@ impl StreamState {
         self.finished = true;
 
         let mut events = Vec::new();
+        self.close_thinking(&mut events);
         if self.text_started && !self.text_finished {
             self.text_finished = true;
             events.push(StreamEvent::ContentBlockStop(ContentBlockStopEvent {
-                index: 0,
+                index: self.text_block_index(),
             }));
         }
 
+        let tool_index_offset = self.tool_index_offset();
         for state in self.tool_calls.values_mut() {
             if !state.started {
-                if let Some(start_event) = state.start_event()? {
+                if let Some(start_event) = state.start_event(tool_index_offset)? {
                     state.started = true;
                     events.push(StreamEvent::ContentBlockStart(start_event));
-                    if let Some(delta_event) = state.delta_event() {
+                    if let Some(delta_event) = state.delta_event(tool_index_offset) {
                         events.push(StreamEvent::ContentBlockDelta(delta_event));
                     }
                 }
@@ -577,7 +610,7 @@ impl StreamState {
             if state.started && !state.stopped {
                 state.stopped = true;
                 events.push(StreamEvent::ContentBlockStop(ContentBlockStopEvent {
-                    index: state.block_index(),
+                    index: state.block_index(tool_index_offset),
                 }));
             }
         }
@@ -603,6 +636,31 @@ impl StreamState {
         }
         Ok(events)
     }
+
+    fn close_thinking(&mut self, events: &mut Vec<StreamEvent>) {
+        if self.thinking_started && !self.thinking_finished {
+            self.thinking_finished = true;
+            events.push(StreamEvent::ContentBlockStop(ContentBlockStopEvent {
+                index: 0,
+            }));
+        }
+    }
+
+    const fn text_block_index(&self) -> u32 {
+        if self.thinking_started {
+            1
+        } else {
+            0
+        }
+    }
+
+    const fn tool_index_offset(&self) -> u32 {
+        if self.thinking_started {
+            2
+        } else {
+            1
+        }
+    }
 }
 
 #[derive(Debug, Default)]
@@ -630,12 +688,12 @@ impl ToolCallState {
         }
     }
 
-    const fn block_index(&self) -> u32 {
+    const fn block_index(&self, offset: u32) -> u32 {
-        self.openai_index + 1
+        self.openai_index + offset
     }
 
     #[allow(clippy::unnecessary_wraps)]
-    fn start_event(&self) -> Result<Option<ContentBlockStartEvent>, ApiError> {
+    fn start_event(&self, offset: u32) -> Result<Option<ContentBlockStartEvent>, ApiError> {
         let Some(name) = self.name.clone() else {
             return Ok(None);
         };
@@ -644,7 +702,7 @@ impl ToolCallState {
             .clone()
             .unwrap_or_else(|| format!("tool_call_{}", self.openai_index));
         Ok(Some(ContentBlockStartEvent {
-            index: self.block_index(),
+            index: self.block_index(offset),
             content_block: OutputContentBlock::ToolUse {
                 id,
                 name,
@@ -653,14 +711,14 @@ impl ToolCallState {
         }))
     }
 
-    fn delta_event(&mut self) -> Option<ContentBlockDeltaEvent> {
+    fn delta_event(&mut self, offset: u32) -> Option<ContentBlockDeltaEvent> {
         if self.emitted_len >= self.arguments.len() {
             return None;
         }
         let delta = self.arguments[self.emitted_len..].to_string();
         self.emitted_len = self.arguments.len();
         Some(ContentBlockDeltaEvent {
-            index: self.block_index(),
+            index: self.block_index(offset),
             delta: ContentBlockDelta::InputJsonDelta {
                 partial_json: delta,
             },
@@ -690,6 +748,8 @@ struct ChatMessage {
     #[serde(default)]
     content: Option<String>,
     #[serde(default)]
+    reasoning_content: Option<String>,
+    #[serde(default)]
     tool_calls: Vec<ResponseToolCall>,
 }
 
@@ -735,6 +795,8 @@ struct ChunkChoice {
 struct ChunkDelta {
     #[serde(default)]
     content: Option<String>,
+    #[serde(default)]
+    reasoning_content: Option<String>,
     #[serde(default, deserialize_with = "deserialize_null_as_empty_vec")]
     tool_calls: Vec<DeltaToolCall>,
 }
@@ -793,6 +855,15 @@ pub fn is_reasoning_model(model: &str) -> bool {
         || canonical.contains("thinking")
 }
 
+/// Returns true for OpenAI-compatible DeepSeek V4 models that require prior
+/// assistant reasoning to be echoed back as `reasoning_content` in history.
+#[must_use]
+pub fn model_requires_reasoning_content_in_history(model: &str) -> bool {
+    let lowered = model.to_ascii_lowercase();
+    let canonical = lowered.rsplit('/').next().unwrap_or(lowered.as_str());
+    canonical.starts_with("deepseek-v4")
+}
+
 /// Strip routing prefix (e.g., "openai/gpt-4" → "gpt-4") for the wire.
 /// The prefix is used only to select transport; the backend expects the
 /// bare model id.
@@ -948,10 +1019,14 @@ pub fn translate_message(message: &InputMessage, model: &str) -> Vec<Value> {
     match message.role.as_str() {
         "assistant" => {
             let mut text = String::new();
+            let mut reasoning = String::new();
             let mut tool_calls = Vec::new();
             for block in &message.content {
                 match block {
                     InputContentBlock::Text { text: value } => text.push_str(value),
+                    InputContentBlock::Thinking {
+                        thinking: value, ..
+                    } => reasoning.push_str(value),
                     InputContentBlock::ToolUse { id, name, input } => tool_calls.push(json!({
                         "id": id,
                         "type": "function",
@@ -963,13 +1038,18 @@ pub fn translate_message(message: &InputMessage, model: &str) -> Vec<Value> {
                     InputContentBlock::ToolResult { .. } => {}
                 }
             }
-            if text.is_empty() && tool_calls.is_empty() {
+            let include_reasoning =
+                model_requires_reasoning_content_in_history(model) && !reasoning.is_empty();
+            if text.is_empty() && tool_calls.is_empty() && !include_reasoning {
                 Vec::new()
             } else {
                 let mut msg = serde_json::json!({
                     "role": "assistant",
                     "content": (!text.is_empty()).then_some(text),
                 });
+                if include_reasoning {
+                    msg["reasoning_content"] = json!(reasoning);
+                }
                 // Only include tool_calls when non-empty: some providers reject
                 // assistant messages with an explicit empty tool_calls array.
                 if !tool_calls.is_empty() {
@@ -1003,6 +1083,7 @@ pub fn translate_message(message: &InputMessage, model: &str) -> Vec<Value> {
                     }
                     Some(msg)
                 }
+                InputContentBlock::Thinking { .. } => None,
                 InputContentBlock::ToolUse { .. } => None,
             })
             .collect(),
@@ -1182,6 +1263,16 @@ fn normalize_response(
             "chat completion response missing choices",
         ))?;
     let mut content = Vec::new();
+    if let Some(thinking) = choice
+        .message
+        .reasoning_content
+        .filter(|value| !value.is_empty())
+    {
+        content.push(OutputContentBlock::Thinking {
+            thinking,
+            signature: None,
+        });
+    }
     if let Some(text) = choice.message.content.filter(|value| !value.is_empty()) {
         content.push(OutputContentBlock::Text { text });
     }
@@ -1413,13 +1504,15 @@ impl StringExt for String {
 mod tests {
     use super::{
         build_chat_completion_request, chat_completions_endpoint, is_reasoning_model,
-        normalize_finish_reason, openai_tool_choice, parse_tool_arguments, OpenAiCompatClient,
+        model_requires_reasoning_content_in_history, normalize_finish_reason, normalize_response,
-        OpenAiCompatConfig,
+        openai_tool_choice, parse_tool_arguments, OpenAiCompatClient, OpenAiCompatConfig,
+        StreamState,
     };
     use crate::error::ApiError;
     use crate::types::{
-        InputContentBlock, InputMessage, MessageRequest, ToolChoice, ToolDefinition,
+        ContentBlockDelta, ContentBlockDeltaEvent, ContentBlockStartEvent, ContentBlockStopEvent,
-        ToolResultContentBlock,
+        InputContentBlock, InputMessage, MessageRequest, OutputContentBlock, StreamEvent,
+        ToolChoice, ToolDefinition, ToolResultContentBlock,
     };
     use serde_json::json;
     use std::sync::{Mutex, OnceLock};
@@ -1465,6 +1558,188 @@ mod tests {
         assert_eq!(payload["tool_choice"], json!("auto"));
     }
 
+    #[test]
+    fn model_requires_reasoning_content_in_history_detects_deepseek_v4_models() {
+        // Given DeepSeek V4 and non-V4 model names.
+        let positive = [
+            "deepseek-v4-flash",
+            "deepseek-v4-pro",
+            "openai/deepseek-v4-pro",
+            "deepseek/deepseek-v4-flash",
+        ];
+        let negative = [
+            "deepseek-reasoner",
+            "deepseek-chat",
+            "gpt-4o",
+            "claude-sonnet-4-6",
+        ];
+
+        // When checking whether history reasoning_content is required.
+        // Then only DeepSeek V4 variants require it.
+        for model in positive {
+            assert!(model_requires_reasoning_content_in_history(model));
+        }
+        for model in negative {
+            assert!(!model_requires_reasoning_content_in_history(model));
+        }
+    }
+
+    #[test]
+    fn legacy_deepseek_reasoner_request_omits_reasoning_content_for_assistant_history() {
+        // Given an assistant history turn containing thinking.
+        let request = assistant_history_with_thinking_request("deepseek-reasoner");
+
+        // When serializing for legacy deepseek-reasoner.
+        let payload = build_chat_completion_request(&request, OpenAiCompatConfig::openai());
+
+        // Then reasoning_content is omitted.
+        let assistant = &payload["messages"][0];
+        assert_eq!(assistant["role"], json!("assistant"));
+        assert!(assistant.get("reasoning_content").is_none());
+    }
+
+    #[test]
+    fn deepseek_v4_pro_request_includes_reasoning_content_for_assistant_history() {
+        // Given an assistant history turn containing thinking.
+        let request = assistant_history_with_thinking_request("openai/deepseek-v4-pro");
+
+        // When serializing for DeepSeek V4 Pro.
+        let payload = build_chat_completion_request(&request, OpenAiCompatConfig::openai());
+
+        // Then reasoning_content is included on the assistant message.
+        let assistant = &payload["messages"][0];
+        assert_eq!(assistant["reasoning_content"], json!("prior reasoning"));
+        assert_eq!(assistant["content"], json!("answer"));
+    }
+
+    #[test]
+    fn deepseek_v4_flash_request_includes_reasoning_content_for_assistant_history() {
+        // Given an assistant history turn containing thinking.
+        let request = assistant_history_with_thinking_request("deepseek-v4-flash");
+
+        // When serializing for DeepSeek V4 Flash.
+        let payload = build_chat_completion_request(&request, OpenAiCompatConfig::openai());
+
+        // Then reasoning_content is included on the assistant message.
+        let assistant = &payload["messages"][0];
+        assert_eq!(assistant["reasoning_content"], json!("prior reasoning"));
+    }
+
+    #[test]
+    fn non_streaming_response_with_reasoning_content_emits_thinking_block_first() {
+        // Given a non-streaming OpenAI-compatible response with reasoning_content.
+        let response = super::ChatCompletionResponse {
+            id: "chatcmpl_reasoning".to_string(),
+            model: "deepseek-v4-pro".to_string(),
+            choices: vec![super::ChatChoice {
+                message: super::ChatMessage {
+                    role: "assistant".to_string(),
+                    content: Some("final answer".to_string()),
+                    reasoning_content: Some("hidden thought".to_string()),
+                    tool_calls: Vec::new(),
+                },
+                finish_reason: Some("stop".to_string()),
+            }],
+            usage: None,
+        };
+
+        // When normalizing the provider response.
+        let normalized = normalize_response("deepseek-v4-pro", response).expect("normalized");
+
+        // Then Thinking is the first content block, before text.
+        assert_eq!(
+            normalized.content,
+            vec![
+                OutputContentBlock::Thinking {
+                    thinking: "hidden thought".to_string(),
+                    signature: None,
+                },
+                OutputContentBlock::Text {
+                    text: "final answer".to_string(),
+                },
+            ]
+        );
+    }
+
+    #[test]
+    fn streaming_chunks_with_reasoning_content_emit_thinking_block_events_before_text() {
+        // Given streaming chunks with reasoning_content followed by text.
+        let mut state = StreamState::new("deepseek-v4-pro".to_string());
+        let mut events = state
+            .ingest_chunk(super::ChatCompletionChunk {
+                id: "chatcmpl_stream_reasoning".to_string(),
+                model: Some("deepseek-v4-pro".to_string()),
+                choices: vec![super::ChunkChoice {
+                    delta: super::ChunkDelta {
+                        content: None,
+                        reasoning_content: Some("think".to_string()),
+                        tool_calls: Vec::new(),
+                    },
+                    finish_reason: None,
+                }],
+                usage: None,
+            })
+            .expect("reasoning chunk");
+        events.extend(
+            state
+                .ingest_chunk(super::ChatCompletionChunk {
+                    id: "chatcmpl_stream_reasoning".to_string(),
+                    model: None,
+                    choices: vec![super::ChunkChoice {
+                        delta: super::ChunkDelta {
+                            content: Some(" answer".to_string()),
+                            reasoning_content: None,
+                            tool_calls: Vec::new(),
+                        },
+                        finish_reason: Some("stop".to_string()),
+                    }],
+                    usage: None,
+                })
+                .expect("text chunk"),
+        );
+        events.extend(state.finish().expect("finish"));
+
+        // When reading normalized stream events.
+        // Then Thinking starts at index 0, text is offset to index 1.
+        assert!(matches!(events[0], StreamEvent::MessageStart(_)));
+        assert!(matches!(
+            events[1],
+            StreamEvent::ContentBlockStart(ContentBlockStartEvent {
+                index: 0,
+                content_block: OutputContentBlock::Thinking { .. },
+            })
+        ));
+        assert!(matches!(
+            events[2],
+            StreamEvent::ContentBlockDelta(ContentBlockDeltaEvent {
+                index: 0,
+                delta: ContentBlockDelta::ThinkingDelta { .. },
+            })
+        ));
+        assert!(matches!(
+            events[3],
+            StreamEvent::ContentBlockStop(ContentBlockStopEvent { index: 0 })
+        ));
+        assert!(matches!(
+            events[4],
+            StreamEvent::ContentBlockStart(ContentBlockStartEvent {
+                index: 1,
+                content_block: OutputContentBlock::Text { .. },
+            })
+        ));
+        assert!(matches!(
+            events[5],
+            StreamEvent::ContentBlockDelta(ContentBlockDeltaEvent {
+                index: 1,
+                delta: ContentBlockDelta::TextDelta { .. },
+            })
+        ));
+        assert!(matches!(
+            events[6],
+            StreamEvent::ContentBlockStop(ContentBlockStopEvent { index: 1 })
+        ));
+    }
+
     #[test]
     fn tool_schema_object_gets_strict_fields_for_responses_endpoint() {
         // OpenAI /responses endpoint rejects object schemas missing
@@ -1624,6 +1899,27 @@ mod tests {
         );
     }
 
+    fn assistant_history_with_thinking_request(model: &str) -> MessageRequest {
+        MessageRequest {
+            model: model.to_string(),
+            max_tokens: 100,
+            messages: vec![InputMessage {
+                role: "assistant".to_string(),
+                content: vec![
+                    InputContentBlock::Thinking {
+                        thinking: "prior reasoning".to_string(),
+                        signature: None,
+                    },
+                    InputContentBlock::Text {
+                        text: "answer".to_string(),
+                    },
+                ],
+            }],
+            stream: false,
+            ..Default::default()
+        }
+    }
+
     fn env_lock() -> std::sync::MutexGuard<'static, ()> {
         static LOCK: OnceLock<Mutex<()>> = OnceLock::new();
         LOCK.get_or_init(|| Mutex::new(()))
@@ -2195,9 +2491,16 @@ mod tests {
 
     #[test]
     fn provider_specific_size_limits_are_correct() {
-        assert_eq!(OpenAiCompatConfig::dashscope().max_request_body_bytes, 6_291_456); // 6MB
+        assert_eq!(
-        assert_eq!(OpenAiCompatConfig::openai().max_request_body_bytes, 104_857_600); // 100MB
+            OpenAiCompatConfig::dashscope().max_request_body_bytes,
-        assert_eq!(OpenAiCompatConfig::xai().max_request_body_bytes, 52_428_800); // 50MB
+            6_291_456
+        ); // 6MB
+        assert_eq!(
+            OpenAiCompatConfig::openai().max_request_body_bytes,
+            104_857_600
+        ); // 100MB
+        assert_eq!(OpenAiCompatConfig::xai().max_request_body_bytes, 52_428_800);
+        // 50MB
     }
 
     #[test]

rust/crates/api/src/types.rs

@@ -81,6 +81,11 @@ pub enum InputContentBlock {
     Text {
         text: String,
     },
+    Thinking {
+        thinking: String,
+        #[serde(default, skip_serializing_if = "Option::is_none")]
+        signature: Option<String>,
+    },
     ToolUse {
         id: String,
         name: String,
@@ -268,8 +273,9 @@ pub enum StreamEvent {
 #[cfg(test)]
 mod tests {
     use runtime::format_usd;
+    use serde_json::json;
 
-    use super::{MessageResponse, Usage};
+    use super::{InputContentBlock, MessageResponse, Usage};
 
     #[test]
     fn usage_total_tokens_includes_cache_tokens() {
@@ -307,4 +313,33 @@ mod tests {
         assert_eq!(format_usd(cost.total_cost_usd()), "$54.6750");
         assert_eq!(response.total_tokens(), 1_800_000);
     }
+
+    #[test]
+    fn input_content_block_thinking_serializes_with_snake_case_type() {
+        // given
+        let block = InputContentBlock::Thinking {
+            thinking: "pondering".to_string(),
+            signature: Some("sig_123".to_string()),
+        };
+
+        // when
+        let serialized = serde_json::to_value(&block).unwrap();
+        let deserialized: InputContentBlock = serde_json::from_value(json!({
+            "type": "thinking",
+            "thinking": "pondering",
+            "signature": "sig_123"
+        }))
+        .unwrap();
+
+        // then
+        assert_eq!(
+            serialized,
+            json!({
+                "type": "thinking",
+                "thinking": "pondering",
+                "signature": "sig_123"
+            })
+        );
+        assert_eq!(deserialized, block);
+    }
 }

rust/crates/api/tests/openai_compat_integration.rs

@@ -63,6 +63,50 @@ async fn send_message_uses_openai_compatible_endpoint_and_auth() {
     assert_eq!(body["tools"][0]["type"], json!("function"));
 }
 
+#[tokio::test]
+async fn send_message_preserves_deepseek_reasoning_content_before_text() {
+    let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));
+    let body = concat!(
+        "{",
+        "\"id\":\"chatcmpl_deepseek_reasoning\",",
+        "\"model\":\"deepseek-v4-pro\",",
+        "\"choices\":[{",
+        "\"message\":{\"role\":\"assistant\",\"reasoning_content\":\"Think first\",\"content\":\"Answer second\",\"tool_calls\":[]},",
+        "\"finish_reason\":\"stop\"",
+        "}],",
+        "\"usage\":{\"prompt_tokens\":11,\"completion_tokens\":5}",
+        "}"
+    );
+    let server = spawn_server(
+        state.clone(),
+        vec![http_response("200 OK", "application/json", body)],
+    )
+    .await;
+
+    let client = OpenAiCompatClient::new("openai-test-key", OpenAiCompatConfig::openai())
+        .with_base_url(server.base_url());
+    let response = client
+        .send_message(&MessageRequest {
+            model: "openai/deepseek-v4-pro".to_string(),
+            ..sample_request(false)
+        })
+        .await
+        .expect("request should succeed");
+
+    assert_eq!(
+        response.content,
+        vec![
+            OutputContentBlock::Thinking {
+                thinking: "Think first".to_string(),
+                signature: None,
+            },
+            OutputContentBlock::Text {
+                text: "Answer second".to_string(),
+            },
+        ]
+    );
+}
+
 #[tokio::test]
 async fn send_message_blocks_oversized_xai_requests_before_the_http_call() {
     let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));

rust/crates/commands/src/lib.rs

@@ -2371,6 +2371,40 @@ pub fn handle_skills_slash_command(args: Option<&str>, cwd: &Path) -> std::io::R
             let skills = load_skills_from_roots(&roots)?;
             Ok(render_skills_report(&skills))
         }
+        Some(args) if args.starts_with("list ") => {
+            let filter = args["list ".len()..].trim().to_lowercase();
+            let roots = discover_skill_roots(cwd);
+            let skills = load_skills_from_roots(&roots)?;
+            let filtered: Vec<_> = skills
+                .into_iter()
+                .filter(|s| s.name.to_lowercase().contains(&filter))
+                .collect();
+            Ok(render_skills_report(&filtered))
+        }
+        Some("show" | "info" | "describe") => {
+            let roots = discover_skill_roots(cwd);
+            let skills = load_skills_from_roots(&roots)?;
+            Ok(render_skills_report(&skills))
+        }
+        Some(args)
+            if args.starts_with("show ")
+                || args.starts_with("info ")
+                || args.starts_with("describe ") =>
+        {
+            let name = args
+                .splitn(2, ' ')
+                .nth(1)
+                .unwrap_or_default()
+                .trim()
+                .to_lowercase();
+            let roots = discover_skill_roots(cwd);
+            let skills = load_skills_from_roots(&roots)?;
+            let matched: Vec<_> = skills
+                .into_iter()
+                .filter(|s| s.name.to_lowercase() == name)
+                .collect();
+            Ok(render_skills_report(&matched))
+        }
         Some("install") => Ok(render_skills_usage(Some("install"))),
         Some(args) if args.starts_with("install ") => {
             let target = args["install ".len()..].trim();
@@ -2402,6 +2436,40 @@ pub fn handle_skills_slash_command_json(args: Option<&str>, cwd: &Path) -> std::
             let skills = load_skills_from_roots(&roots)?;
             Ok(render_skills_report_json(&skills))
         }
+        Some(args) if args.starts_with("list ") => {
+            let filter = args["list ".len()..].trim().to_lowercase();
+            let roots = discover_skill_roots(cwd);
+            let skills = load_skills_from_roots(&roots)?;
+            let filtered: Vec<_> = skills
+                .into_iter()
+                .filter(|s| s.name.to_lowercase().contains(&filter))
+                .collect();
+            Ok(render_skills_report_json(&filtered))
+        }
+        Some("show" | "info" | "describe") => {
+            let roots = discover_skill_roots(cwd);
+            let skills = load_skills_from_roots(&roots)?;
+            Ok(render_skills_report_json(&skills))
+        }
+        Some(args)
+            if args.starts_with("show ")
+                || args.starts_with("info ")
+                || args.starts_with("describe ") =>
+        {
+            let name = args
+                .splitn(2, ' ')
+                .nth(1)
+                .unwrap_or_default()
+                .trim()
+                .to_lowercase();
+            let roots = discover_skill_roots(cwd);
+            let skills = load_skills_from_roots(&roots)?;
+            let matched: Vec<_> = skills
+                .into_iter()
+                .filter(|s| s.name.to_lowercase() == name)
+                .collect();
+            Ok(render_skills_report_json(&matched))
+        }
         Some("install") => Ok(render_skills_usage_json(Some("install"))),
         Some(args) if args.starts_with("install ") => {
             let target = args["install ".len()..].trim();
@@ -2419,10 +2487,27 @@ pub fn handle_skills_slash_command_json(args: Option<&str>, cwd: &Path) -> std::
 #[must_use]
 pub fn classify_skills_slash_command(args: Option<&str>) -> SkillSlashDispatch {
     match normalize_optional_args(args) {
-        None | Some("list" | "help" | "-h" | "--help") => SkillSlashDispatch::Local,
+        None | Some("list" | "help" | "-h" | "--help" | "show" | "info" | "describe") => {
+            SkillSlashDispatch::Local
+        }
+        Some(args)
+            if args
+                .split_whitespace()
+                .any(|part| matches!(part, "-h" | "--help")) =>
+        {
+            SkillSlashDispatch::Local
+        }
         Some(args) if args == "install" || args.starts_with("install ") => {
             SkillSlashDispatch::Local
         }
+        Some(args)
+            if args.starts_with("list ")
+                || args.starts_with("show ")
+                || args.starts_with("info ")
+                || args.starts_with("describe ") =>
+        {
+            SkillSlashDispatch::Local
+        }
         Some(args) => SkillSlashDispatch::Invoke(format!("${}", args.trim_start_matches('/'))),
     }
 }
@@ -2596,10 +2681,44 @@ fn render_mcp_report_for(
                 )),
             }
         }
+        Some(args) if args.split_whitespace().next() == Some("list") && args.contains(' ') => {
+            // `mcp list <filter>` — list does not accept arguments; treat as unsupported action.
+            Ok(render_mcp_unsupported_action_text(
+                args,
+                "list accepts no filter argument; use `claw mcp list`",
+            ))
+        }
+        Some(args) if matches!(args.split_whitespace().next(), Some("info" | "describe")) => {
+            Ok(render_mcp_unsupported_action_text(
+                args,
+                "use `claw mcp show <server>` to inspect a server",
+            ))
+        }
         Some(args) => Ok(render_mcp_usage(Some(args))),
     }
 }
 
+fn render_mcp_unsupported_action_text(action: &str, hint: &str) -> String {
+    format!(
+        "MCP\n  Error            unsupported action '{action}'\n  Hint             {hint}\n  Usage            /mcp [list|show <server>|help]"
+    )
+}
+
+fn render_mcp_unsupported_action_json(action: &str, hint: &str) -> Value {
+    json!({
+        "kind": "mcp",
+        "action": "error",
+        "ok": false,
+        "error_kind": "unsupported_action",
+        "requested_action": action,
+        "hint": hint,
+        "usage": {
+            "slash_command": "/mcp [list|show <server>|help]",
+            "direct_cli": "claw mcp [list|show <server>|help]",
+        },
+    })
+}
+
 fn render_mcp_report_json_for(
     loader: &ConfigLoader,
     cwd: &Path,
@@ -2623,10 +2742,8 @@ fn render_mcp_report_json_for(
             // runs, the existing serializer adds `status: "ok"` below.
             match loader.load() {
                 Ok(runtime_config) => {
-                    let mut value = render_mcp_summary_report_json(
+                    let mut value =
-                        cwd,
+                        render_mcp_summary_report_json(cwd, runtime_config.mcp().servers());
-                        runtime_config.mcp().servers(),
-                    );
                     if let Some(map) = value.as_object_mut() {
                         map.insert("status".to_string(), Value::String("ok".to_string()));
                         map.insert("config_load_error".to_string(), Value::Null);
@@ -2682,6 +2799,18 @@ fn render_mcp_report_json_for(
                 })),
             }
         }
+        Some(args) if args.split_whitespace().next() == Some("list") && args.contains(' ') => {
+            Ok(render_mcp_unsupported_action_json(
+                args,
+                "list accepts no filter argument; use `claw mcp list`",
+            ))
+        }
+        Some(args) if matches!(args.split_whitespace().next(), Some("info" | "describe")) => {
+            Ok(render_mcp_unsupported_action_json(
+                args,
+                "use `claw mcp show <server>` to inspect a server",
+            ))
+        }
         Some(args) => Ok(render_mcp_usage_json(Some(args))),
     }
 }
@@ -4621,6 +4750,32 @@ mod tests {
         assert!(agents_error.contains("  Usage            /agents [list|help]"));
     }
 
+    #[test]
+    fn skills_show_and_list_filter_do_not_invoke_model() {
+        // `show`, `info`, `list <filter>` must route to Local, not Invoke.
+        // Regression for: `claw skills show plan` unexpectedly spawned a model session.
+        for token in &["show", "info", "describe"] {
+            assert_eq!(
+                classify_skills_slash_command(Some(token)),
+                SkillSlashDispatch::Local,
+                "`skills {token}` alone must be Local"
+            );
+        }
+        for prefix in &["show ", "info ", "list ", "describe "] {
+            let arg = format!("{prefix}plan");
+            assert_eq!(
+                classify_skills_slash_command(Some(&arg)),
+                SkillSlashDispatch::Local,
+                "`skills {arg}` must be Local, not Invoke"
+            );
+        }
+        // Bare invocable tokens still dispatch to Invoke.
+        assert_eq!(
+            classify_skills_slash_command(Some("plan")),
+            SkillSlashDispatch::Invoke("$plan".to_string()),
+        );
+    }
+
     #[test]
     fn accepts_skills_invocation_arguments_for_prompt_dispatch() {
         assert_eq!(
@@ -4643,6 +4798,38 @@ mod tests {
         );
     }
 
+    #[test]
+    fn mcp_unsupported_actions_return_typed_error_not_generic_help() {
+        // `mcp info <name>` and `mcp list <filter>` must return typed errors, not raw help.
+        // Regression for #504: these previously fell through to render_mcp_usage with
+        // unexpected=arg, giving no machine-readable error_kind.
+        use crate::handle_mcp_slash_command_json;
+        use std::path::PathBuf;
+        let cwd = PathBuf::from("/tmp");
+
+        let info_json = handle_mcp_slash_command_json(Some("info nonexistent"), &cwd)
+            .expect("info nonexistent should not error at IO level");
+        assert_eq!(info_json["kind"], "mcp");
+        assert_eq!(info_json["ok"], false);
+        assert_eq!(info_json["error_kind"], "unsupported_action");
+        assert!(info_json["hint"]
+            .as_str()
+            .unwrap_or_default()
+            .contains("show"));
+
+        let list_filter_json = handle_mcp_slash_command_json(Some("list nonexistent"), &cwd)
+            .expect("list nonexistent should not error at IO level");
+        assert_eq!(list_filter_json["kind"], "mcp");
+        assert_eq!(list_filter_json["ok"], false);
+        assert_eq!(list_filter_json["error_kind"], "unsupported_action");
+
+        let describe_json = handle_mcp_slash_command_json(Some("describe myserver"), &cwd)
+            .expect("describe myserver should not error at IO level");
+        assert_eq!(describe_json["kind"], "mcp");
+        assert_eq!(describe_json["ok"], false);
+        assert_eq!(describe_json["error_kind"], "unsupported_action");
+    }
+
     #[test]
     fn rejects_invalid_mcp_arguments() {
         let show_error = parse_error_message("/mcp show alpha beta");

rust/crates/mock-anthropic-service/src/lib.rs

@@ -248,6 +248,7 @@ fn detect_scenario(request: &MessageRequest) -> Option<Scenario> {
                 .split_whitespace()
                 .find_map(|token| token.strip_prefix(SCENARIO_PREFIX))
                 .and_then(Scenario::parse),
+            InputContentBlock::Thinking { .. } => None,
             _ => None,
         })
     })

rust/crates/runtime/src/approval_tokens.rs

@@ -0,0 +1,502 @@
+use std::collections::BTreeMap;
+
+/// Machine-readable policy exception scope that an approval token may override.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ApprovalScope {
+    pub policy: String,
+    pub action: String,
+    pub repository: Option<String>,
+    pub branch: Option<String>,
+}
+
+impl ApprovalScope {
+    #[must_use]
+    pub fn new(policy: impl Into<String>, action: impl Into<String>) -> Self {
+        Self {
+            policy: policy.into(),
+            action: action.into(),
+            repository: None,
+            branch: None,
+        }
+    }
+
+    #[must_use]
+    pub fn with_repository(mut self, repository: impl Into<String>) -> Self {
+        self.repository = Some(repository.into());
+        self
+    }
+
+    #[must_use]
+    pub fn with_branch(mut self, branch: impl Into<String>) -> Self {
+        self.branch = Some(branch.into());
+        self
+    }
+}
+
+/// Actor/session hop recorded when an approval is delegated or consumed.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ApprovalDelegationHop {
+    pub actor: String,
+    pub session_id: Option<String>,
+    pub reason: String,
+}
+
+impl ApprovalDelegationHop {
+    #[must_use]
+    pub fn new(actor: impl Into<String>, reason: impl Into<String>) -> Self {
+        Self {
+            actor: actor.into(),
+            session_id: None,
+            reason: reason.into(),
+        }
+    }
+
+    #[must_use]
+    pub fn with_session_id(mut self, session_id: impl Into<String>) -> Self {
+        self.session_id = Some(session_id.into());
+        self
+    }
+}
+
+/// Current lifecycle state for a policy-exception approval token.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum ApprovalTokenStatus {
+    Pending,
+    Granted,
+    Consumed,
+    Expired,
+    Revoked,
+}
+
+impl ApprovalTokenStatus {
+    #[must_use]
+    pub fn as_str(self) -> &'static str {
+        match self {
+            Self::Pending => "approval_pending",
+            Self::Granted => "approval_granted",
+            Self::Consumed => "approval_consumed",
+            Self::Expired => "approval_expired",
+            Self::Revoked => "approval_revoked",
+        }
+    }
+}
+
+/// Typed policy errors returned when a token cannot authorize a blocked action.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum ApprovalTokenError {
+    NoApproval,
+    ApprovalPending,
+    ApprovalExpired,
+    ApprovalRevoked,
+    ApprovalAlreadyConsumed,
+    ScopeMismatch {
+        expected: Box<ApprovalScope>,
+        actual: Box<ApprovalScope>,
+    },
+    UnauthorizedDelegate {
+        expected: String,
+        actual: String,
+    },
+}
+
+impl ApprovalTokenError {
+    #[must_use]
+    pub fn as_str(&self) -> &'static str {
+        match self {
+            Self::NoApproval => "no_approval",
+            Self::ApprovalPending => "approval_pending",
+            Self::ApprovalExpired => "approval_expired",
+            Self::ApprovalRevoked => "approval_revoked",
+            Self::ApprovalAlreadyConsumed => "approval_already_consumed",
+            Self::ScopeMismatch { .. } => "approval_scope_mismatch",
+            Self::UnauthorizedDelegate { .. } => "approval_unauthorized_delegate",
+        }
+    }
+}
+
+/// Approval grant bound to a policy/action scope, approving owner, and executor.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ApprovalTokenGrant {
+    pub token: String,
+    pub scope: ApprovalScope,
+    pub approving_actor: String,
+    pub approved_executor: String,
+    pub status: ApprovalTokenStatus,
+    pub expires_at_epoch_seconds: Option<u64>,
+    pub max_uses: u32,
+    pub uses: u32,
+    delegation_chain: Vec<ApprovalDelegationHop>,
+}
+
+impl ApprovalTokenGrant {
+    #[must_use]
+    pub fn pending(
+        token: impl Into<String>,
+        scope: ApprovalScope,
+        approving_actor: impl Into<String>,
+        approved_executor: impl Into<String>,
+    ) -> Self {
+        Self {
+            token: token.into(),
+            scope,
+            approving_actor: approving_actor.into(),
+            approved_executor: approved_executor.into(),
+            status: ApprovalTokenStatus::Pending,
+            expires_at_epoch_seconds: None,
+            max_uses: 1,
+            uses: 0,
+            delegation_chain: Vec::new(),
+        }
+    }
+
+    #[must_use]
+    pub fn granted(
+        token: impl Into<String>,
+        scope: ApprovalScope,
+        approving_actor: impl Into<String>,
+        approved_executor: impl Into<String>,
+    ) -> Self {
+        Self::pending(token, scope, approving_actor, approved_executor).approve()
+    }
+
+    #[must_use]
+    pub fn approve(mut self) -> Self {
+        self.status = ApprovalTokenStatus::Granted;
+        self
+    }
+
+    #[must_use]
+    pub fn expires_at(mut self, epoch_seconds: u64) -> Self {
+        self.expires_at_epoch_seconds = Some(epoch_seconds);
+        self
+    }
+
+    #[must_use]
+    pub fn with_max_uses(mut self, max_uses: u32) -> Self {
+        self.max_uses = max_uses.max(1);
+        self
+    }
+
+    #[must_use]
+    pub fn with_delegation_hop(mut self, hop: ApprovalDelegationHop) -> Self {
+        self.delegation_chain.push(hop);
+        self
+    }
+
+    #[must_use]
+    pub fn delegation_chain(&self) -> &[ApprovalDelegationHop] {
+        &self.delegation_chain
+    }
+}
+
+/// Auditable result of verifying or consuming an approval token.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct ApprovalTokenAudit {
+    pub token: String,
+    pub scope: ApprovalScope,
+    pub approving_actor: String,
+    pub executing_actor: String,
+    pub status: ApprovalTokenStatus,
+    pub delegated_execution: bool,
+    pub delegation_chain: Vec<ApprovalDelegationHop>,
+    pub uses: u32,
+    pub max_uses: u32,
+}
+
+/// In-memory approval-token ledger with one-time-use and replay protection.
+#[derive(Debug, Clone, PartialEq, Eq, Default)]
+pub struct ApprovalTokenLedger {
+    grants: BTreeMap<String, ApprovalTokenGrant>,
+}
+
+impl ApprovalTokenLedger {
+    #[must_use]
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    pub fn insert(&mut self, grant: ApprovalTokenGrant) {
+        self.grants.insert(grant.token.clone(), grant);
+    }
+
+    #[must_use]
+    pub fn get(&self, token: &str) -> Option<&ApprovalTokenGrant> {
+        self.grants.get(token)
+    }
+
+    pub fn revoke(&mut self, token: &str) -> Result<ApprovalTokenAudit, ApprovalTokenError> {
+        let grant = self
+            .grants
+            .get_mut(token)
+            .ok_or(ApprovalTokenError::NoApproval)?;
+        grant.status = ApprovalTokenStatus::Revoked;
+        Ok(Self::audit_for(grant, &grant.approved_executor))
+    }
+
+    pub fn verify(
+        &self,
+        token: &str,
+        scope: &ApprovalScope,
+        executing_actor: &str,
+        now_epoch_seconds: u64,
+    ) -> Result<ApprovalTokenAudit, ApprovalTokenError> {
+        let grant = self
+            .grants
+            .get(token)
+            .ok_or(ApprovalTokenError::NoApproval)?;
+        Self::validate_grant(grant, scope, executing_actor, now_epoch_seconds)?;
+        Ok(Self::audit_for(grant, executing_actor))
+    }
+
+    pub fn consume(
+        &mut self,
+        token: &str,
+        scope: &ApprovalScope,
+        executing_actor: &str,
+        now_epoch_seconds: u64,
+    ) -> Result<ApprovalTokenAudit, ApprovalTokenError> {
+        let grant = self
+            .grants
+            .get_mut(token)
+            .ok_or(ApprovalTokenError::NoApproval)?;
+        Self::validate_grant(grant, scope, executing_actor, now_epoch_seconds)?;
+        grant.uses += 1;
+        if grant.uses >= grant.max_uses {
+            grant.status = ApprovalTokenStatus::Consumed;
+        }
+        Ok(Self::audit_for(grant, executing_actor))
+    }
+
+    fn validate_grant(
+        grant: &ApprovalTokenGrant,
+        scope: &ApprovalScope,
+        executing_actor: &str,
+        now_epoch_seconds: u64,
+    ) -> Result<(), ApprovalTokenError> {
+        match grant.status {
+            ApprovalTokenStatus::Pending => return Err(ApprovalTokenError::ApprovalPending),
+            ApprovalTokenStatus::Consumed => {
+                return Err(ApprovalTokenError::ApprovalAlreadyConsumed)
+            }
+            ApprovalTokenStatus::Expired => return Err(ApprovalTokenError::ApprovalExpired),
+            ApprovalTokenStatus::Revoked => return Err(ApprovalTokenError::ApprovalRevoked),
+            ApprovalTokenStatus::Granted => {}
+        }
+
+        if grant
+            .expires_at_epoch_seconds
+            .is_some_and(|expires_at| now_epoch_seconds > expires_at)
+        {
+            return Err(ApprovalTokenError::ApprovalExpired);
+        }
+
+        if grant.uses >= grant.max_uses {
+            return Err(ApprovalTokenError::ApprovalAlreadyConsumed);
+        }
+
+        if grant.scope != *scope {
+            return Err(ApprovalTokenError::ScopeMismatch {
+                expected: Box::new(grant.scope.clone()),
+                actual: Box::new(scope.clone()),
+            });
+        }
+
+        if grant.approved_executor != executing_actor {
+            return Err(ApprovalTokenError::UnauthorizedDelegate {
+                expected: grant.approved_executor.clone(),
+                actual: executing_actor.to_string(),
+            });
+        }
+
+        Ok(())
+    }
+
+    fn audit_for(grant: &ApprovalTokenGrant, executing_actor: &str) -> ApprovalTokenAudit {
+        let mut delegation_chain = grant.delegation_chain.clone();
+        if delegation_chain.is_empty() {
+            delegation_chain.push(ApprovalDelegationHop::new(
+                grant.approving_actor.clone(),
+                "approval granted",
+            ));
+        }
+        if grant.approving_actor != executing_actor
+            && !delegation_chain
+                .iter()
+                .any(|hop| hop.actor == executing_actor)
+        {
+            delegation_chain.push(ApprovalDelegationHop::new(
+                executing_actor.to_string(),
+                "delegated execution",
+            ));
+        }
+
+        ApprovalTokenAudit {
+            token: grant.token.clone(),
+            scope: grant.scope.clone(),
+            approving_actor: grant.approving_actor.clone(),
+            executing_actor: executing_actor.to_string(),
+            status: grant.status,
+            delegated_execution: grant.approving_actor != executing_actor,
+            delegation_chain,
+            uses: grant.uses,
+            max_uses: grant.max_uses,
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{
+        ApprovalDelegationHop, ApprovalScope, ApprovalTokenError, ApprovalTokenGrant,
+        ApprovalTokenLedger, ApprovalTokenStatus,
+    };
+
+    #[test]
+    fn approval_token_blocks_until_owner_grants_policy_exception() {
+        let mut ledger = ApprovalTokenLedger::new();
+        let scope = ApprovalScope::new("main_push_forbidden", "git push")
+            .with_repository("sisyphus/claw-code")
+            .with_branch("main");
+        ledger.insert(ApprovalTokenGrant::pending(
+            "tok-pending",
+            scope.clone(),
+            "repo-owner",
+            "release-bot",
+        ));
+
+        assert!(matches!(
+            ledger.verify("tok-missing", &scope, "release-bot", 10),
+            Err(ApprovalTokenError::NoApproval)
+        ));
+        assert!(matches!(
+            ledger.verify("tok-pending", &scope, "release-bot", 10),
+            Err(ApprovalTokenError::ApprovalPending)
+        ));
+
+        ledger.insert(ApprovalTokenGrant::granted(
+            "tok-granted",
+            scope.clone(),
+            "repo-owner",
+            "release-bot",
+        ));
+        let audit = ledger
+            .verify("tok-granted", &scope, "release-bot", 10)
+            .expect("owner approval should verify");
+
+        assert_eq!(audit.status, ApprovalTokenStatus::Granted);
+        assert_eq!(audit.approving_actor, "repo-owner");
+        assert_eq!(audit.executing_actor, "release-bot");
+        assert!(audit.delegated_execution);
+    }
+
+    #[test]
+    fn approval_token_is_one_time_use_and_rejects_replay() {
+        let mut ledger = ApprovalTokenLedger::new();
+        let scope = ApprovalScope::new("release_requires_owner", "release publish")
+            .with_repository("sisyphus/claw-code");
+        ledger.insert(ApprovalTokenGrant::granted(
+            "tok-once",
+            scope.clone(),
+            "owner",
+            "release-bot",
+        ));
+
+        let first = ledger
+            .consume("tok-once", &scope, "release-bot", 10)
+            .expect("first use should consume token");
+        assert_eq!(first.status, ApprovalTokenStatus::Consumed);
+        assert_eq!(first.uses, 1);
+
+        assert!(matches!(
+            ledger.consume("tok-once", &scope, "release-bot", 11),
+            Err(ApprovalTokenError::ApprovalAlreadyConsumed)
+        ));
+        assert_eq!(
+            ledger.get("tok-once").map(|grant| grant.status),
+            Some(ApprovalTokenStatus::Consumed)
+        );
+    }
+
+    #[test]
+    fn approval_token_rejects_scope_expansion_expiry_and_revocation() {
+        let mut ledger = ApprovalTokenLedger::new();
+        let scope = ApprovalScope::new("main_push_forbidden", "git push")
+            .with_repository("sisyphus/claw-code")
+            .with_branch("main");
+        let dev_scope = ApprovalScope::new("main_push_forbidden", "git push")
+            .with_repository("sisyphus/claw-code")
+            .with_branch("dev");
+
+        ledger.insert(
+            ApprovalTokenGrant::granted("tok-expiring", scope.clone(), "owner", "bot")
+                .expires_at(20),
+        );
+
+        assert!(matches!(
+            ledger.verify("tok-expiring", &dev_scope, "bot", 10),
+            Err(ApprovalTokenError::ScopeMismatch { .. })
+        ));
+        assert!(matches!(
+            ledger.verify("tok-expiring", &scope, "bot", 21),
+            Err(ApprovalTokenError::ApprovalExpired)
+        ));
+
+        ledger.insert(ApprovalTokenGrant::granted(
+            "tok-revoked",
+            scope.clone(),
+            "owner",
+            "bot",
+        ));
+        let revoked = ledger
+            .revoke("tok-revoked")
+            .expect("revocation should be audited");
+        assert_eq!(revoked.status, ApprovalTokenStatus::Revoked);
+        assert!(matches!(
+            ledger.verify("tok-revoked", &scope, "bot", 10),
+            Err(ApprovalTokenError::ApprovalRevoked)
+        ));
+    }
+
+    #[test]
+    fn approval_token_preserves_delegation_traceability() {
+        let mut ledger = ApprovalTokenLedger::new();
+        let scope = ApprovalScope::new("deploy_requires_owner", "deploy prod");
+        ledger.insert(
+            ApprovalTokenGrant::granted("tok-delegated", scope.clone(), "owner", "deploy-bot")
+                .with_delegation_hop(
+                    ApprovalDelegationHop::new("owner", "owner approval")
+                        .with_session_id("session-owner"),
+                )
+                .with_delegation_hop(
+                    ApprovalDelegationHop::new("lead-agent", "handoff to deploy bot")
+                        .with_session_id("session-lead"),
+                ),
+        );
+
+        assert!(matches!(
+            ledger.verify("tok-delegated", &scope, "unexpected-bot", 10),
+            Err(ApprovalTokenError::UnauthorizedDelegate { expected, actual })
+                if expected == "deploy-bot" && actual == "unexpected-bot"
+        ));
+
+        let audit = ledger
+            .consume("tok-delegated", &scope, "deploy-bot", 10)
+            .expect("approved delegate should consume token");
+        let actors = audit
+            .delegation_chain
+            .iter()
+            .map(|hop| hop.actor.as_str())
+            .collect::<Vec<_>>();
+
+        assert!(audit.delegated_execution);
+        assert_eq!(actors, vec!["owner", "lead-agent", "deploy-bot"]);
+        assert_eq!(
+            audit.delegation_chain[0].session_id.as_deref(),
+            Some("session-owner")
+        );
+        assert_eq!(
+            audit.delegation_chain[1].session_id.as_deref(),
+            Some("session-lead")
+        );
+    }
+}

rust/crates/runtime/src/bash.rs

@@ -4,6 +4,7 @@ use std::process::{Command, Stdio};
 use std::time::Duration;
 
 use serde::{Deserialize, Serialize};
+use serde_json::json;
 use tokio::process::Command as TokioCommand;
 use tokio::runtime::Builder;
 use tokio::time::timeout;
@@ -122,7 +123,7 @@ fn detect_and_emit_ship_prepared(command: &str) {
             actor: get_git_actor().unwrap_or_else(|| "unknown".to_string()),
             pr_number: None,
         };
-        let _event = LaneEvent::ship_prepared(format!("{}", now), &provenance);
+        let _event = LaneEvent::ship_prepared(format!("{now}"), &provenance);
         // Log to stderr as interim routing before event stream integration
         eprintln!(
             "[ship.prepared] branch={} -> main, commits={}, actor={}",
@@ -172,31 +173,14 @@ async fn execute_bash_async(
 ) -> io::Result<BashCommandOutput> {
     // Detect and emit ship provenance for git push operations
     detect_and_emit_ship_prepared(&input.command);
-    
+
     let mut command = prepare_tokio_command(&input.command, &cwd, &sandbox_status, true);
 
     let output_result = if let Some(timeout_ms) = input.timeout {
-        match timeout(Duration::from_millis(timeout_ms), command.output()).await {
+        if let Ok(result) = timeout(Duration::from_millis(timeout_ms), command.output()).await {
-            Ok(result) => (result?, false),
+            (result?, false)
-            Err(_) => {
+        } else {
-                return Ok(BashCommandOutput {
+            return Ok(timeout_output(&input, timeout_ms, sandbox_status));
-                    stdout: String::new(),
-                    stderr: format!("Command exceeded timeout of {timeout_ms} ms"),
-                    raw_output_path: None,
-                    interrupted: true,
-                    is_image: None,
-                    background_task_id: None,
-                    backgrounded_by_user: None,
-                    assistant_auto_backgrounded: None,
-                    dangerously_disable_sandbox: input.dangerously_disable_sandbox,
-                    return_code_interpretation: Some(String::from("timeout")),
-                    no_output_expected: Some(true),
-                    structured_content: None,
-                    persisted_output_path: None,
-                    persisted_output_size: None,
-                    sandbox_status: Some(sandbox_status),
-                });
-            }
         }
     } else {
         (command.output().await?, false)
@@ -233,6 +217,67 @@ async fn execute_bash_async(
     })
 }
 
+fn timeout_output(
+    input: &BashCommandInput,
+    timeout_ms: u64,
+    sandbox_status: SandboxStatus,
+) -> BashCommandOutput {
+    let is_test = is_test_command(&input.command);
+    let return_code_interpretation = if is_test { "test.hung" } else { "timeout" };
+    BashCommandOutput {
+        stdout: String::new(),
+        stderr: format!("Command exceeded timeout of {timeout_ms} ms"),
+        raw_output_path: None,
+        interrupted: true,
+        is_image: None,
+        background_task_id: None,
+        backgrounded_by_user: None,
+        assistant_auto_backgrounded: None,
+        dangerously_disable_sandbox: input.dangerously_disable_sandbox,
+        return_code_interpretation: Some(String::from(return_code_interpretation)),
+        no_output_expected: Some(true),
+        structured_content: Some(vec![test_timeout_provenance(
+            &input.command,
+            timeout_ms,
+            is_test,
+        )]),
+        persisted_output_path: None,
+        persisted_output_size: None,
+        sandbox_status: Some(sandbox_status),
+    }
+}
+
+fn is_test_command(command: &str) -> bool {
+    let normalized = command
+        .split_whitespace()
+        .collect::<Vec<_>>()
+        .join(" ")
+        .to_ascii_lowercase();
+    normalized.contains("cargo test")
+        || normalized.contains("cargo nextest")
+        || normalized.contains("npm test")
+        || normalized.contains("pnpm test")
+        || normalized.contains("yarn test")
+        || normalized.contains("pytest")
+}
+
+fn test_timeout_provenance(
+    command: &str,
+    timeout_ms: u64,
+    classified_as_test_hang: bool,
+) -> serde_json::Value {
+    json!({
+        "event": if classified_as_test_hang { "test.hung" } else { "command.timeout" },
+        "failureClass": if classified_as_test_hang { "test_hang" } else { "timeout" },
+        "data": {
+            "command": command,
+            "timeoutMs": timeout_ms,
+            "provenance": "bash.timeout",
+            "classification": if classified_as_test_hang { "test.hung" } else { "timeout" }
+        }
+    })
+}
+
 fn sandbox_status_for_input(input: &BashCommandInput, cwd: &std::path::Path) -> SandboxStatus {
     let config = ConfigLoader::default_for(cwd).load().map_or_else(
         |_| SandboxConfig::default(),
@@ -349,6 +394,31 @@ mod tests {
 
         assert!(!output.sandbox_status.expect("sandbox status").enabled);
     }
+
+    #[test]
+    fn timed_out_test_command_is_classified_as_hung_test_with_provenance() {
+        let output = execute_bash(BashCommandInput {
+            command: String::from("sleep 1 # cargo test slow_case"),
+            timeout: Some(1),
+            description: None,
+            run_in_background: Some(false),
+            dangerously_disable_sandbox: Some(false),
+            namespace_restrictions: Some(false),
+            isolate_network: Some(false),
+            filesystem_mode: Some(FilesystemIsolationMode::WorkspaceOnly),
+            allowed_mounts: None,
+        })
+        .expect("bash command should return structured timeout");
+
+        assert!(output.interrupted);
+        assert_eq!(
+            output.return_code_interpretation.as_deref(),
+            Some("test.hung")
+        );
+        let structured = output.structured_content.expect("structured content");
+        assert_eq!(structured[0]["event"], "test.hung");
+        assert_eq!(structured[0]["data"]["provenance"], "bash.timeout");
+    }
 }
 
 /// Maximum output bytes before truncation (16 KiB, matching upstream).

rust/crates/runtime/src/compact.rs

@@ -212,7 +212,7 @@ fn summarize_messages(messages: &[ConversationMessage]) -> String {
         .filter_map(|block| match block {
             ContentBlock::ToolUse { name, .. } => Some(name.as_str()),
             ContentBlock::ToolResult { tool_name, .. } => Some(tool_name.as_str()),
-            ContentBlock::Text { .. } => None,
+            ContentBlock::Text { .. } | ContentBlock::Thinking { .. } => None,
         })
         .collect::<Vec<_>>();
     tool_names.sort_unstable();
@@ -317,6 +317,9 @@ fn merge_compact_summaries(existing_summary: Option<&str>, new_summary: &str) ->
 fn summarize_block(block: &ContentBlock) -> String {
     let raw = match block {
         ContentBlock::Text { text } => text.clone(),
+        ContentBlock::Thinking { thinking, .. } => {
+            format!("thinking ({} chars)", thinking.chars().count())
+        }
         ContentBlock::ToolUse { name, input, .. } => format!("tool_use {name}({input})"),
         ContentBlock::ToolResult {
             tool_name,
@@ -378,6 +381,7 @@ fn collect_key_files(messages: &[ConversationMessage]) -> Vec<String> {
             ContentBlock::Text { text } => text.as_str(),
             ContentBlock::ToolUse { input, .. } => input.as_str(),
             ContentBlock::ToolResult { output, .. } => output.as_str(),
+            ContentBlock::Thinking { thinking, .. } => thinking.as_str(),
         })
         .flat_map(extract_file_candidates)
         .collect::<Vec<_>>();
@@ -400,6 +404,7 @@ fn first_text_block(message: &ConversationMessage) -> Option<&str> {
         ContentBlock::Text { text } if !text.trim().is_empty() => Some(text.as_str()),
         ContentBlock::ToolUse { .. }
         | ContentBlock::ToolResult { .. }
+        | ContentBlock::Thinking { .. }
         | ContentBlock::Text { .. } => None,
     })
 }
@@ -450,6 +455,10 @@ fn estimate_message_tokens(message: &ConversationMessage) -> usize {
             ContentBlock::ToolResult {
                 tool_name, output, ..
             } => (tool_name.len() + output.len()) / 4 + 1,
+            ContentBlock::Thinking {
+                thinking,
+                signature,
+            } => thinking.len() / 4 + signature.as_ref().map_or(0, |value| value.len() / 4 + 1),
         })
         .sum()
 }

rust/crates/runtime/src/config.rs

@@ -414,6 +414,17 @@ impl RuntimeConfig {
     pub fn trusted_roots(&self) -> &[String] {
         &self.feature_config.trusted_roots
     }
+
+    /// Merge config-level default trusted roots with per-call roots.
+    ///
+    /// Config roots are defaults and are kept first; per-call roots extend the
+    /// allowlist for a specific worker/session creation request. Duplicates are
+    /// removed without reordering the first occurrence so evidence remains
+    /// deterministic while avoiding repeated trust checks.
+    #[must_use]
+    pub fn trusted_roots_with_overrides(&self, per_call_roots: &[String]) -> Vec<String> {
+        merge_trusted_roots(self.trusted_roots(), per_call_roots)
+    }
 }
 
 impl RuntimeFeatureConfig {
@@ -483,6 +494,22 @@ impl RuntimeFeatureConfig {
     pub fn trusted_roots(&self) -> &[String] {
         &self.trusted_roots
     }
+
+    /// Merge this config's default trusted roots with per-call roots.
+    #[must_use]
+    pub fn trusted_roots_with_overrides(&self, per_call_roots: &[String]) -> Vec<String> {
+        merge_trusted_roots(self.trusted_roots(), per_call_roots)
+    }
+}
+
+fn merge_trusted_roots(config_roots: &[String], per_call_roots: &[String]) -> Vec<String> {
+    let mut merged = Vec::with_capacity(config_roots.len() + per_call_roots.len());
+    for root in config_roots.iter().chain(per_call_roots.iter()) {
+        if !merged.contains(root) {
+            merged.push(root.clone());
+        }
+    }
+    merged
 }
 
 impl ProviderFallbackConfig {
@@ -1245,8 +1272,8 @@ fn push_unique(target: &mut Vec<String>, value: String) {
 mod tests {
     use super::{
         deep_merge_objects, parse_permission_mode_label, ConfigLoader, ConfigSource,
-        McpServerConfig, McpTransport, ResolvedPermissionMode, RuntimeHookConfig,
+        McpServerConfig, McpTransport, ResolvedPermissionMode, RuntimeFeatureConfig,
-        RuntimePluginConfig, CLAW_SETTINGS_SCHEMA_NAME,
+        RuntimeHookConfig, RuntimePluginConfig, CLAW_SETTINGS_SCHEMA_NAME,
     };
     use crate::json::JsonValue;
     use crate::sandbox::FilesystemIsolationMode;
@@ -1502,6 +1529,51 @@ mod tests {
         fs::remove_dir_all(root).expect("cleanup temp dir");
     }
 
+    #[test]
+    fn trusted_roots_with_overrides_preserves_config_defaults_and_adds_per_call_roots() {
+        // given
+        let root = temp_dir();
+        let cwd = root.join("project");
+        let home = root.join("home").join(".claw");
+        fs::create_dir_all(&home).expect("home config dir");
+        fs::create_dir_all(&cwd).expect("project dir");
+        fs::write(
+            home.join("settings.json"),
+            r#"{"trustedRoots": ["/tmp/config-default", "/tmp/shared"]}"#,
+        )
+        .expect("write settings");
+
+        // when
+        let loaded = ConfigLoader::new(&cwd, &home)
+            .load()
+            .expect("config should load");
+        let merged = loaded.trusted_roots_with_overrides(&[
+            "/tmp/per-call".to_string(),
+            "/tmp/shared".to_string(),
+        ]);
+
+        // then
+        assert_eq!(
+            merged,
+            ["/tmp/config-default", "/tmp/shared", "/tmp/per-call"]
+        );
+
+        fs::remove_dir_all(root).expect("cleanup temp dir");
+    }
+
+    #[test]
+    fn runtime_feature_trusted_roots_with_overrides_matches_runtime_config_merge() {
+        let config = RuntimeFeatureConfig {
+            trusted_roots: vec!["/tmp/config".to_string()],
+            ..RuntimeFeatureConfig::default()
+        };
+
+        assert_eq!(
+            config.trusted_roots_with_overrides(&["/tmp/per-call".to_string()]),
+            ["/tmp/config", "/tmp/per-call"]
+        );
+    }
+
     #[test]
     fn trusted_roots_default_is_empty_when_unset() {
         // given

rust/crates/runtime/src/conversation.rs

@@ -28,6 +28,10 @@ pub struct ApiRequest {
 /// Streamed events emitted while processing a single assistant turn.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub enum AssistantEvent {
+    Thinking {
+        thinking: String,
+        signature: Option<String>,
+    },
     TextDelta(String),
     ToolUse {
         id: String,
@@ -721,6 +725,16 @@ fn build_assistant_message(
 
     for event in events {
         match event {
+            AssistantEvent::Thinking {
+                thinking,
+                signature,
+            } => {
+                flush_text_block(&mut text, &mut blocks);
+                blocks.push(ContentBlock::Thinking {
+                    thinking,
+                    signature,
+                });
+            }
             AssistantEvent::TextDelta(delta) => text.push_str(&delta),
             AssistantEvent::ToolUse { id, name, input } => {
                 flush_text_block(&mut text, &mut blocks);
@@ -1723,6 +1737,47 @@ mod tests {
             .contains("assistant stream produced no content"));
     }
 
+    #[test]
+    fn build_assistant_message_places_thinking_block_before_text_and_tool_use() {
+        // given
+        let events = vec![
+            AssistantEvent::Thinking {
+                thinking: "pondering".to_string(),
+                signature: Some("sig".to_string()),
+            },
+            AssistantEvent::TextDelta("hello".to_string()),
+            AssistantEvent::ToolUse {
+                id: "tool-1".to_string(),
+                name: "echo".to_string(),
+                input: "payload".to_string(),
+            },
+            AssistantEvent::MessageStop,
+        ];
+
+        // when
+        let (message, _, _) = build_assistant_message(events)
+            .expect("assistant message should preserve thinking, text, and tool blocks");
+
+        // then
+        assert_eq!(
+            message.blocks,
+            vec![
+                ContentBlock::Thinking {
+                    thinking: "pondering".to_string(),
+                    signature: Some("sig".to_string()),
+                },
+                ContentBlock::Text {
+                    text: "hello".to_string(),
+                },
+                ContentBlock::ToolUse {
+                    id: "tool-1".to_string(),
+                    name: "echo".to_string(),
+                    input: "payload".to_string(),
+                },
+            ]
+        );
+    }
+
     #[test]
     fn static_tool_executor_rejects_unknown_tools() {
         // given

rust/crates/runtime/src/file_ops.rs

@@ -1,4 +1,5 @@
 use std::cmp::Reverse;
+use std::collections::HashSet;
 use std::fs;
 use std::io;
 use std::path::{Path, PathBuf};
@@ -7,7 +8,7 @@ use std::time::Instant;
 use glob::Pattern;
 use regex::RegexBuilder;
 use serde::{Deserialize, Serialize};
-use walkdir::WalkDir;
+use walkdir::{DirEntry, WalkDir};
 
 /// Maximum file size that can be read (10 MB).
 const MAX_READ_SIZE: u64 = 10 * 1024 * 1024;
@@ -15,6 +16,15 @@ const MAX_READ_SIZE: u64 = 10 * 1024 * 1024;
 /// Maximum file size that can be written (10 MB).
 const MAX_WRITE_SIZE: usize = 10 * 1024 * 1024;
 
+const GLOB_SEARCH_IGNORED_DIRS: &[&str] = &[
+    ".git",
+    "node_modules",
+    ".build",
+    "target",
+    "dist",
+    "coverage",
+];
+
 /// Check whether a file appears to contain binary content by examining
 /// the first chunk for NUL bytes.
 fn is_binary_file(path: &Path) -> io::Result<bool> {
@@ -297,11 +307,23 @@ pub fn edit_file(
 
 /// Expands a glob pattern and returns matching filenames.
 pub fn glob_search(pattern: &str, path: Option<&str>) -> io::Result<GlobSearchOutput> {
+    glob_search_impl(pattern, path, None)
+}
+
+fn glob_search_impl(
+    pattern: &str,
+    path: Option<&str>,
+    workspace_root: Option<&Path>,
+) -> io::Result<GlobSearchOutput> {
     let started = Instant::now();
     let base_dir = path
         .map(normalize_path)
         .transpose()?
         .unwrap_or(std::env::current_dir()?);
+    let canonical_root = workspace_root.map(canonicalize_workspace_root);
+    if let Some(root) = canonical_root.as_deref() {
+        validate_workspace_boundary(&base_dir, root)?;
+    }
     let search_pattern = if Path::new(pattern).is_absolute() {
         pattern.to_owned()
     } else {
@@ -313,14 +335,32 @@ pub fn glob_search(pattern: &str, path: Option<&str>) -> io::Result<GlobSearchOu
     // `Assets/**/*.{cs,uxml,uss}` work correctly.
     let expanded = expand_braces(&search_pattern);
 
-    let mut seen = std::collections::HashSet::new();
+    let mut seen = HashSet::new();
     let mut matches = Vec::new();
     for pat in &expanded {
-        let entries = glob::glob(pat)
+        let compiled = Pattern::new(pat)
             .map_err(|error| io::Error::new(io::ErrorKind::InvalidInput, error.to_string()))?;
+        let walk_root = derive_glob_walk_root(pat);
+        if let Some(root) = canonical_root.as_deref() {
+            let canonical_walk_root = walk_root
+                .canonicalize()
+                .unwrap_or_else(|_| walk_root.clone());
+            validate_workspace_boundary(&canonical_walk_root, root)?;
+        }
+        let entries = WalkDir::new(&walk_root)
+            .into_iter()
+            .filter_entry(|entry| !should_skip_glob_dir(entry));
         for entry in entries.flatten() {
-            if entry.is_file() && seen.insert(entry.clone()) {
+            let candidate = entry.path();
-                matches.push(entry);
+            if entry.file_type().is_file()
+                && compiled.matches_path(candidate)
+                && seen.insert(candidate.to_path_buf())
+            {
+                if let Some(root) = canonical_root.as_deref() {
+                    let canonical_candidate = candidate.canonicalize()?;
+                    validate_workspace_boundary(&canonical_candidate, root)?;
+                }
+                matches.push(candidate.to_path_buf());
             }
         }
     }
@@ -349,12 +389,23 @@ pub fn glob_search(pattern: &str, path: Option<&str>) -> io::Result<GlobSearchOu
 
 /// Runs a regex search over workspace files with optional context lines.
 pub fn grep_search(input: &GrepSearchInput) -> io::Result<GrepSearchOutput> {
+    grep_search_impl(input, None)
+}
+
+fn grep_search_impl(
+    input: &GrepSearchInput,
+    workspace_root: Option<&Path>,
+) -> io::Result<GrepSearchOutput> {
     let base_path = input
         .path
         .as_deref()
         .map(normalize_path)
         .transpose()?
         .unwrap_or(std::env::current_dir()?);
+    let canonical_root = workspace_root.map(canonicalize_workspace_root);
+    if let Some(root) = canonical_root.as_deref() {
+        validate_workspace_boundary(&base_path, root)?;
+    }
 
     let regex = RegexBuilder::new(&input.pattern)
         .case_insensitive(input.case_insensitive.unwrap_or(false))
@@ -380,6 +431,10 @@ pub fn grep_search(input: &GrepSearchInput) -> io::Result<GrepSearchOutput> {
     let mut total_matches = 0usize;
 
     for file_path in collect_search_files(&base_path)? {
+        if let Some(root) = canonical_root.as_deref() {
+            let canonical_file = file_path.canonicalize()?;
+            validate_workspace_boundary(&canonical_file, root)?;
+        }
         if !matches_optional_filters(&file_path, glob_filter.as_ref(), file_type) {
             continue;
         }
@@ -429,27 +484,21 @@ pub fn grep_search(input: &GrepSearchInput) -> io::Result<GrepSearchOutput> {
 
     let (filenames, applied_limit, applied_offset) =
         apply_limit(filenames, input.head_limit, input.offset);
-    let content_output = if output_mode == "content" {
+    if output_mode == "content" {
-        let (lines, limit, offset) = apply_limit(content_lines, input.head_limit, input.offset);
+        return Ok(build_grep_content_output(
-        return Ok(GrepSearchOutput {
+            output_mode,
-            mode: Some(output_mode),
-            num_files: filenames.len(),
             filenames,
-            num_lines: Some(lines.len()),
+            content_lines,
-            content: Some(lines.join("\n")),
+            input.head_limit,
-            num_matches: None,
+            input.offset,
-            applied_limit: limit,
+        ));
-            applied_offset: offset,
+    }
-        });
-    } else {
-        None
-    };
 
     Ok(GrepSearchOutput {
         mode: Some(output_mode.clone()),
         num_files: filenames.len(),
         filenames,
-        content: content_output,
+        content: None,
         num_lines: None,
         num_matches: (output_mode == "count").then_some(total_matches),
         applied_limit,
@@ -457,6 +506,65 @@ pub fn grep_search(input: &GrepSearchInput) -> io::Result<GrepSearchOutput> {
     })
 }
 
+fn build_grep_content_output(
+    output_mode: String,
+    filenames: Vec<String>,
+    content_lines: Vec<String>,
+    head_limit: Option<usize>,
+    offset: Option<usize>,
+) -> GrepSearchOutput {
+    let (lines, limit, offset) = apply_limit(content_lines, head_limit, offset);
+    GrepSearchOutput {
+        mode: Some(output_mode),
+        num_files: filenames.len(),
+        filenames,
+        num_lines: Some(lines.len()),
+        content: Some(lines.join("\n")),
+        num_matches: None,
+        applied_limit: limit,
+        applied_offset: offset,
+    }
+}
+
+fn canonicalize_workspace_root(workspace_root: &Path) -> PathBuf {
+    workspace_root
+        .canonicalize()
+        .unwrap_or_else(|_| workspace_root.to_path_buf())
+}
+
+fn should_skip_glob_dir(entry: &DirEntry) -> bool {
+    entry.file_type().is_dir()
+        && entry
+            .file_name()
+            .to_str()
+            .is_some_and(|name| GLOB_SEARCH_IGNORED_DIRS.contains(&name))
+}
+
+fn derive_glob_walk_root(pattern: &str) -> PathBuf {
+    let path = Path::new(pattern);
+    let mut prefix = PathBuf::new();
+    let mut saw_component = false;
+
+    for component in path.components() {
+        let text = component.as_os_str().to_string_lossy();
+        if component_contains_glob(&text) {
+            break;
+        }
+        prefix.push(component.as_os_str());
+        saw_component = true;
+    }
+
+    if saw_component {
+        prefix
+    } else {
+        std::env::current_dir().unwrap_or_else(|_| PathBuf::from("."))
+    }
+}
+
+fn component_contains_glob(component: &str) -> bool {
+    component.contains('*') || component.contains('?') || component.contains('[')
+}
+
 fn collect_search_files(base_path: &Path) -> io::Result<Vec<PathBuf>> {
     if base_path.is_file() {
         return Ok(vec![base_path.to_path_buf()]);
@@ -574,9 +682,7 @@ pub fn read_file_in_workspace(
     workspace_root: &Path,
 ) -> io::Result<ReadFileOutput> {
     let absolute_path = normalize_path(path)?;
-    let canonical_root = workspace_root
+    let canonical_root = canonicalize_workspace_root(workspace_root);
-        .canonicalize()
-        .unwrap_or_else(|_| workspace_root.to_path_buf());
     validate_workspace_boundary(&absolute_path, &canonical_root)?;
     read_file(path, offset, limit)
 }
@@ -589,9 +695,7 @@ pub fn write_file_in_workspace(
     workspace_root: &Path,
 ) -> io::Result<WriteFileOutput> {
     let absolute_path = normalize_path_allow_missing(path)?;
-    let canonical_root = workspace_root
+    let canonical_root = canonicalize_workspace_root(workspace_root);
-        .canonicalize()
-        .unwrap_or_else(|_| workspace_root.to_path_buf());
     validate_workspace_boundary(&absolute_path, &canonical_root)?;
     write_file(path, content)
 }
@@ -606,13 +710,30 @@ pub fn edit_file_in_workspace(
     workspace_root: &Path,
 ) -> io::Result<EditFileOutput> {
     let absolute_path = normalize_path(path)?;
-    let canonical_root = workspace_root
+    let canonical_root = canonicalize_workspace_root(workspace_root);
-        .canonicalize()
-        .unwrap_or_else(|_| workspace_root.to_path_buf());
     validate_workspace_boundary(&absolute_path, &canonical_root)?;
     edit_file(path, old_string, new_string, replace_all)
 }
 
+/// Expand a glob pattern with workspace boundary enforcement.
+#[allow(dead_code)]
+pub fn glob_search_in_workspace(
+    pattern: &str,
+    path: Option<&str>,
+    workspace_root: &Path,
+) -> io::Result<GlobSearchOutput> {
+    glob_search_impl(pattern, path, Some(workspace_root))
+}
+
+/// Search file contents with workspace boundary enforcement.
+#[allow(dead_code)]
+pub fn grep_search_in_workspace(
+    input: &GrepSearchInput,
+    workspace_root: &Path,
+) -> io::Result<GrepSearchOutput> {
+    grep_search_impl(input, Some(workspace_root))
+}
+
 /// Check whether a path is a symlink that resolves outside the workspace.
 #[allow(dead_code)]
 pub fn is_symlink_escape(path: &Path, workspace_root: &Path) -> io::Result<bool> {
@@ -651,11 +772,13 @@ fn expand_braces(pattern: &str) -> Vec<String> {
 
 #[cfg(test)]
 mod tests {
+    use std::path::PathBuf;
     use std::time::{SystemTime, UNIX_EPOCH};
 
     use super::{
-        edit_file, expand_braces, glob_search, grep_search, is_symlink_escape, read_file,
+        component_contains_glob, derive_glob_walk_root, edit_file, expand_braces, glob_search,
-        read_file_in_workspace, write_file, GrepSearchInput, MAX_WRITE_SIZE,
+        grep_search, is_symlink_escape, read_file, read_file_in_workspace, write_file,
+        write_file_in_workspace, GrepSearchInput, MAX_WRITE_SIZE,
     };
 
     fn temp_path(name: &str) -> std::path::PathBuf {
@@ -755,6 +878,68 @@ mod tests {
         assert!(!is_symlink_escape(&normal, &workspace).expect("check should succeed"));
     }
 
+    #[test]
+    #[cfg(unix)]
+    fn workspace_read_rejects_symlink_escape_regression_3007_class() {
+        let workspace = temp_path("workspace-read-symlink-escape");
+        let outside = temp_path("workspace-read-symlink-target");
+        std::fs::create_dir_all(&workspace).expect("workspace dir should be created");
+        std::fs::create_dir_all(&outside).expect("outside dir should be created");
+        let outside_file = outside.join("secret.txt");
+        std::fs::write(&outside_file, "outside secret").expect("outside file should write");
+
+        let link_path = workspace.join("linked-secret.txt");
+        std::os::unix::fs::symlink(&outside_file, &link_path).expect("symlink should create");
+
+        let result =
+            read_file_in_workspace(link_path.to_string_lossy().as_ref(), None, None, &workspace);
+
+        assert!(result.is_err(), "symlink escape must be rejected");
+        let error = result.unwrap_err();
+        assert_eq!(error.kind(), std::io::ErrorKind::PermissionDenied);
+        assert!(
+            error.to_string().contains("escapes workspace"),
+            "error should explain workspace escape: {error}"
+        );
+
+        let _ = std::fs::remove_dir_all(&workspace);
+        let _ = std::fs::remove_dir_all(&outside);
+    }
+
+    #[test]
+    #[cfg(unix)]
+    fn workspace_write_rejects_parent_symlink_escape_regression_3007_class() {
+        let workspace = temp_path("workspace-write-symlink-escape");
+        let outside = temp_path("workspace-write-symlink-target");
+        std::fs::create_dir_all(&workspace).expect("workspace dir should be created");
+        std::fs::create_dir_all(&outside).expect("outside dir should be created");
+
+        let link_dir = workspace.join("linked-outside");
+        std::os::unix::fs::symlink(&outside, &link_dir).expect("symlink dir should create");
+        let escaped_child = link_dir.join("created.txt");
+
+        let result = write_file_in_workspace(
+            escaped_child.to_string_lossy().as_ref(),
+            "must not escape",
+            &workspace,
+        );
+
+        assert!(result.is_err(), "parent symlink escape must be rejected");
+        let error = result.unwrap_err();
+        assert_eq!(error.kind(), std::io::ErrorKind::PermissionDenied);
+        assert!(
+            error.to_string().contains("escapes workspace"),
+            "error should explain workspace escape: {error}"
+        );
+        assert!(
+            !outside.join("created.txt").exists(),
+            "write should not create through an escaping symlink"
+        );
+
+        let _ = std::fs::remove_dir_all(&workspace);
+        let _ = std::fs::remove_dir_all(&outside);
+    }
+
     #[test]
     fn globs_and_greps_directory() {
         let dir = temp_path("search-dir");
@@ -836,4 +1021,50 @@ mod tests {
         );
         let _ = std::fs::remove_dir_all(&dir);
     }
+
+    #[test]
+    fn glob_search_skips_common_heavy_directories() {
+        let dir = temp_path("glob-ignored-dirs");
+        std::fs::create_dir_all(dir.join("src")).unwrap();
+        std::fs::create_dir_all(dir.join("docs")).unwrap();
+        std::fs::create_dir_all(dir.join("node_modules/pkg")).unwrap();
+        std::fs::create_dir_all(dir.join(".build/checkouts/pkg")).unwrap();
+        std::fs::create_dir_all(dir.join("target/debug/deps")).unwrap();
+
+        std::fs::write(dir.join("src/AGENTS.md"), "src").unwrap();
+        std::fs::write(dir.join("docs/AGENTS.md"), "docs").unwrap();
+        std::fs::write(dir.join("node_modules/pkg/AGENTS.md"), "node_modules").unwrap();
+        std::fs::write(dir.join(".build/checkouts/pkg/AGENTS.md"), ".build").unwrap();
+        std::fs::write(dir.join("target/debug/deps/AGENTS.md"), "target").unwrap();
+
+        let result =
+            glob_search("**/AGENTS.md", Some(dir.to_str().unwrap())).expect("glob should succeed");
+
+        assert_eq!(result.num_files, 2, "ignored dirs should be pruned");
+        assert!(result
+            .filenames
+            .iter()
+            .any(|path| path.ends_with("src/AGENTS.md")));
+        assert!(result
+            .filenames
+            .iter()
+            .any(|path| path.ends_with("docs/AGENTS.md")));
+        assert!(!result
+            .filenames
+            .iter()
+            .any(|path| path.contains("node_modules")
+                || path.contains(".build")
+                || path.contains("/target/")));
+
+        let _ = std::fs::remove_dir_all(&dir);
+    }
+
+    #[test]
+    fn derive_glob_walk_root_stops_at_first_glob_component() {
+        let root = derive_glob_walk_root("/tmp/demo/**/AGENTS.md");
+        assert_eq!(root, PathBuf::from("/tmp/demo"));
+        assert!(component_contains_glob("**"));
+        assert!(component_contains_glob("*.rs"));
+        assert!(!component_contains_glob("src"));
+    }
 }

rust/crates/runtime/src/g004_conformance.rs

@@ -0,0 +1,399 @@
+//! Machine-checkable conformance helpers for G004 event/report contract bundles.
+//!
+//! The harness intentionally validates JSON-shaped artifacts instead of owning the
+//! lane-event, report, or approval-token implementations. This keeps it usable by
+//! independent implementation lanes and by golden fixtures produced outside the
+//! runtime crate.
+
+use serde_json::Value;
+
+const BUNDLE_SCHEMA_VERSION: &str = "g004.contract.bundle.v1";
+const REPORT_SCHEMA_VERSION: &str = "g004.report.v1";
+
+/// A single conformance validation failure.
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub struct G004ConformanceError {
+    /// JSON pointer-ish path to the invalid field.
+    pub path: String,
+    /// Human-readable reason the field failed validation.
+    pub message: String,
+}
+
+impl G004ConformanceError {
+    fn new(path: impl Into<String>, message: impl Into<String>) -> Self {
+        Self {
+            path: path.into(),
+            message: message.into(),
+        }
+    }
+}
+
+/// Validate a G004 golden contract bundle.
+///
+/// The bundle shape is deliberately small and cross-lane:
+/// - `laneEvents[]` must expose stable event identity, ordering/provenance, and
+///   terminal dedupe fingerprints.
+/// - `reports[]` must expose schema identity, content hash, projection/redaction
+///   provenance, capability negotiation, fact/hypothesis/negative-evidence
+///   labels, confidence, and field-level delta attribution.
+/// - `approvalTokens[]` must expose owner/scope, delegation chain, one-time-use,
+///   and replay-prevention fields.
+#[must_use]
+pub fn validate_g004_contract_bundle(bundle: &Value) -> Vec<G004ConformanceError> {
+    let mut errors = Vec::new();
+
+    require_string_eq(bundle, "/schemaVersion", BUNDLE_SCHEMA_VERSION, &mut errors);
+    validate_lane_events(bundle.get("laneEvents"), "/laneEvents", &mut errors);
+    validate_reports(bundle.get("reports"), "/reports", &mut errors);
+    validate_approval_tokens(bundle.get("approvalTokens"), "/approvalTokens", &mut errors);
+
+    errors
+}
+
+#[must_use]
+pub fn is_g004_contract_bundle_valid(bundle: &Value) -> bool {
+    validate_g004_contract_bundle(bundle).is_empty()
+}
+
+fn validate_lane_events(value: Option<&Value>, path: &str, errors: &mut Vec<G004ConformanceError>) {
+    let Some(events) = non_empty_array(value, path, errors) else {
+        return;
+    };
+
+    let mut previous_seq = None;
+    for (index, event) in events.iter().enumerate() {
+        let base = format!("{path}/{index}");
+        require_non_empty_string_at(event, "/event", &format!("{base}/event"), errors);
+        require_non_empty_string_at(event, "/status", &format!("{base}/status"), errors);
+        require_non_empty_string_at(event, "/emittedAt", &format!("{base}/emittedAt"), errors);
+        require_non_empty_string_at(
+            event,
+            "/metadata/provenance",
+            &format!("{base}/metadata/provenance"),
+            errors,
+        );
+        require_non_empty_string_at(
+            event,
+            "/metadata/emitterIdentity",
+            &format!("{base}/metadata/emitterIdentity"),
+            errors,
+        );
+        require_non_empty_string_at(
+            event,
+            "/metadata/environmentLabel",
+            &format!("{base}/metadata/environmentLabel"),
+            errors,
+        );
+
+        match get_path(event, "/metadata/seq").and_then(Value::as_u64) {
+            Some(seq) => {
+                if let Some(previous) = previous_seq {
+                    if seq <= previous {
+                        errors.push(G004ConformanceError::new(
+                            format!("{base}/metadata/seq"),
+                            "sequence must be strictly increasing",
+                        ));
+                    }
+                }
+                previous_seq = Some(seq);
+            }
+            None => errors.push(G004ConformanceError::new(
+                format!("{base}/metadata/seq"),
+                "required u64 field missing",
+            )),
+        }
+
+        if is_terminal_event_value(event.get("event")) {
+            require_non_empty_string_at(
+                event,
+                "/metadata/eventFingerprint",
+                &format!("{base}/metadata/eventFingerprint"),
+                errors,
+            );
+        }
+    }
+}
+
+fn validate_reports(value: Option<&Value>, path: &str, errors: &mut Vec<G004ConformanceError>) {
+    let Some(reports) = non_empty_array(value, path, errors) else {
+        return;
+    };
+
+    for (index, report) in reports.iter().enumerate() {
+        let base = format!("{path}/{index}");
+        require_string_eq_at(
+            report,
+            "/schemaVersion",
+            &format!("{base}/schemaVersion"),
+            REPORT_SCHEMA_VERSION,
+            errors,
+        );
+        require_non_empty_string_at(report, "/reportId", &format!("{base}/reportId"), errors);
+        require_non_empty_string_at(
+            report,
+            "/identity/contentHash",
+            &format!("{base}/identity/contentHash"),
+            errors,
+        );
+        require_non_empty_string_at(
+            report,
+            "/projection/provenance",
+            &format!("{base}/projection/provenance"),
+            errors,
+        );
+        require_non_empty_string_at(
+            report,
+            "/redaction/provenance",
+            &format!("{base}/redaction/provenance"),
+            errors,
+        );
+        non_empty_array(
+            get_path(report, "/consumerCapabilities"),
+            &format!("{base}/consumerCapabilities"),
+            errors,
+        );
+        validate_findings(
+            get_path(report, "/findings"),
+            &format!("{base}/findings"),
+            errors,
+        );
+        validate_field_deltas(
+            get_path(report, "/fieldDeltas"),
+            &format!("{base}/fieldDeltas"),
+            errors,
+        );
+    }
+}
+
+fn validate_findings(value: Option<&Value>, path: &str, errors: &mut Vec<G004ConformanceError>) {
+    let Some(findings) = non_empty_array(value, path, errors) else {
+        return;
+    };
+
+    for (index, finding) in findings.iter().enumerate() {
+        let base = format!("{path}/{index}");
+        require_one_of_at(
+            finding,
+            "/kind",
+            &format!("{base}/kind"),
+            &["fact", "hypothesis", "negative_evidence"],
+            errors,
+        );
+        require_one_of_at(
+            finding,
+            "/confidence",
+            &format!("{base}/confidence"),
+            &["low", "medium", "high"],
+            errors,
+        );
+        require_non_empty_string_at(finding, "/statement", &format!("{base}/statement"), errors);
+    }
+}
+
+fn validate_field_deltas(
+    value: Option<&Value>,
+    path: &str,
+    errors: &mut Vec<G004ConformanceError>,
+) {
+    let Some(deltas) = non_empty_array(value, path, errors) else {
+        return;
+    };
+
+    for (index, delta) in deltas.iter().enumerate() {
+        let base = format!("{path}/{index}");
+        require_non_empty_string_at(delta, "/field", &format!("{base}/field"), errors);
+        require_non_empty_string_at(
+            delta,
+            "/previousHash",
+            &format!("{base}/previousHash"),
+            errors,
+        );
+        require_non_empty_string_at(
+            delta,
+            "/currentHash",
+            &format!("{base}/currentHash"),
+            errors,
+        );
+        require_non_empty_string_at(
+            delta,
+            "/attribution",
+            &format!("{base}/attribution"),
+            errors,
+        );
+    }
+}
+
+fn validate_approval_tokens(
+    value: Option<&Value>,
+    path: &str,
+    errors: &mut Vec<G004ConformanceError>,
+) {
+    let Some(tokens) = non_empty_array(value, path, errors) else {
+        return;
+    };
+
+    for (index, token) in tokens.iter().enumerate() {
+        let base = format!("{path}/{index}");
+        require_non_empty_string_at(token, "/tokenId", &format!("{base}/tokenId"), errors);
+        require_non_empty_string_at(token, "/owner", &format!("{base}/owner"), errors);
+        require_non_empty_string_at(token, "/scope", &format!("{base}/scope"), errors);
+        require_non_empty_string_at(token, "/issuedAt", &format!("{base}/issuedAt"), errors);
+        require_bool_true_at(token, "/oneTimeUse", &format!("{base}/oneTimeUse"), errors);
+        require_non_empty_string_at(
+            token,
+            "/replayPreventionNonce",
+            &format!("{base}/replayPreventionNonce"),
+            errors,
+        );
+        validate_delegation_chain(
+            get_path(token, "/delegationChain"),
+            &format!("{base}/delegationChain"),
+            errors,
+        );
+    }
+}
+
+fn validate_delegation_chain(
+    value: Option<&Value>,
+    path: &str,
+    errors: &mut Vec<G004ConformanceError>,
+) {
+    let Some(chain) = non_empty_array(value, path, errors) else {
+        return;
+    };
+
+    for (index, hop) in chain.iter().enumerate() {
+        let base = format!("{path}/{index}");
+        require_non_empty_string_at(hop, "/from", &format!("{base}/from"), errors);
+        require_non_empty_string_at(hop, "/to", &format!("{base}/to"), errors);
+        require_non_empty_string_at(hop, "/action", &format!("{base}/action"), errors);
+        require_non_empty_string_at(hop, "/at", &format!("{base}/at"), errors);
+    }
+}
+
+fn non_empty_array<'a>(
+    value: Option<&'a Value>,
+    path: &str,
+    errors: &mut Vec<G004ConformanceError>,
+) -> Option<&'a Vec<Value>> {
+    match value.and_then(Value::as_array) {
+        Some(array) if !array.is_empty() => Some(array),
+        Some(_) => {
+            errors.push(G004ConformanceError::new(path, "array must not be empty"));
+            None
+        }
+        None => {
+            errors.push(G004ConformanceError::new(
+                path,
+                "required array field missing",
+            ));
+            None
+        }
+    }
+}
+
+fn require_string_eq(
+    root: &Value,
+    path: &str,
+    expected: &str,
+    errors: &mut Vec<G004ConformanceError>,
+) {
+    require_string_eq_at(root, path, path, expected, errors);
+}
+
+fn require_string_eq_at(
+    root: &Value,
+    pointer: &str,
+    error_path: &str,
+    expected: &str,
+    errors: &mut Vec<G004ConformanceError>,
+) {
+    match get_path(root, pointer).and_then(Value::as_str) {
+        Some(actual) if actual == expected => {}
+        Some(actual) => errors.push(G004ConformanceError::new(
+            error_path,
+            format!("expected '{expected}', got '{actual}'"),
+        )),
+        None => errors.push(G004ConformanceError::new(
+            error_path,
+            "required string field missing",
+        )),
+    }
+}
+
+fn require_non_empty_string_at(
+    root: &Value,
+    pointer: &str,
+    error_path: &str,
+    errors: &mut Vec<G004ConformanceError>,
+) {
+    match get_path(root, pointer).and_then(Value::as_str) {
+        Some(value) if !value.trim().is_empty() => {}
+        Some(_) => errors.push(G004ConformanceError::new(
+            error_path,
+            "string must not be empty",
+        )),
+        None => errors.push(G004ConformanceError::new(
+            error_path,
+            "required string field missing",
+        )),
+    }
+}
+
+fn require_one_of_at(
+    root: &Value,
+    pointer: &str,
+    error_path: &str,
+    allowed: &[&str],
+    errors: &mut Vec<G004ConformanceError>,
+) {
+    match get_path(root, pointer).and_then(Value::as_str) {
+        Some(value) if allowed.contains(&value) => {}
+        Some(value) => errors.push(G004ConformanceError::new(
+            error_path,
+            format!("'{value}' is not one of {}", allowed.join(", ")),
+        )),
+        None => errors.push(G004ConformanceError::new(
+            error_path,
+            "required string field missing",
+        )),
+    }
+}
+
+fn require_bool_true_at(
+    root: &Value,
+    pointer: &str,
+    error_path: &str,
+    errors: &mut Vec<G004ConformanceError>,
+) {
+    match get_path(root, pointer).and_then(Value::as_bool) {
+        Some(true) => {}
+        Some(false) => errors.push(G004ConformanceError::new(error_path, "must be true")),
+        None => errors.push(G004ConformanceError::new(
+            error_path,
+            "required boolean field missing",
+        )),
+    }
+}
+
+fn is_terminal_event_value(value: Option<&Value>) -> bool {
+    matches!(
+        value.and_then(Value::as_str),
+        Some("lane.finished" | "lane.failed" | "lane.merged" | "lane.superseded" | "lane.closed")
+    )
+}
+
+fn get_path<'a>(root: &'a Value, path: &str) -> Option<&'a Value> {
+    if let Some(value) = root.pointer(path) {
+        return Some(value);
+    }
+
+    let segments = path.trim_start_matches('/').split('/').collect::<Vec<_>>();
+    for index in 1..segments.len() {
+        let relative = format!("/{}", segments[index..].join("/"));
+        if let Some(value) = root.pointer(&relative) {
+            return Some(value);
+        }
+    }
+    None
+}

rust/crates/runtime/src/green_contract.rs

@@ -27,19 +27,38 @@ impl std::fmt::Display for GreenLevel {
     }
 }
 
-#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
 pub struct GreenContract {
     pub required_level: GreenLevel,
+    pub requirements: Vec<GreenContractRequirement>,
+    pub block_known_flakes: bool,
 }
 
 impl GreenContract {
     #[must_use]
     pub fn new(required_level: GreenLevel) -> Self {
-        Self { required_level }
+        Self {
+            required_level,
+            requirements: Vec::new(),
+            block_known_flakes: false,
+        }
     }
 
     #[must_use]
-    pub fn evaluate(self, observed_level: Option<GreenLevel>) -> GreenContractOutcome {
+    pub fn merge_ready(required_level: GreenLevel) -> Self {
+        Self {
+            required_level,
+            requirements: vec![
+                GreenContractRequirement::TestCommandProvenance,
+                GreenContractRequirement::BaseBranchFreshness,
+                GreenContractRequirement::RecoveryAttemptContext,
+            ],
+            block_known_flakes: true,
+        }
+    }
+
+    #[must_use]
+    pub fn evaluate(&self, observed_level: Option<GreenLevel>) -> GreenContractOutcome {
         match observed_level {
             Some(level) if level >= self.required_level => GreenContractOutcome::Satisfied {
                 required_level: self.required_level,
@@ -53,11 +72,170 @@ impl GreenContract {
     }
 
     #[must_use]
-    pub fn is_satisfied_by(self, observed_level: GreenLevel) -> bool {
+    pub fn evaluate_evidence(&self, evidence: &GreenEvidence) -> GreenEvidenceOutcome {
+        let mut missing = Vec::new();
+        let mut blocking_flakes = Vec::new();
+
+        if evidence.observed_level < self.required_level {
+            missing.push(GreenContractRequirement::RequiredLevel);
+        }
+
+        for requirement in &self.requirements {
+            match requirement {
+                GreenContractRequirement::TestCommandProvenance
+                    if !evidence.has_passing_test_command() =>
+                {
+                    missing.push(*requirement);
+                }
+                GreenContractRequirement::BaseBranchFreshness if !evidence.base_branch_fresh => {
+                    missing.push(*requirement);
+                }
+                GreenContractRequirement::RecoveryAttemptContext
+                    if !evidence.recovery_attempt_context_recorded =>
+                {
+                    missing.push(*requirement);
+                }
+                _ => {}
+            }
+        }
+
+        if self.block_known_flakes {
+            blocking_flakes = evidence
+                .known_flakes
+                .iter()
+                .filter(|flake| flake.blocks_green)
+                .cloned()
+                .collect();
+        }
+
+        if missing.is_empty() && blocking_flakes.is_empty() {
+            GreenEvidenceOutcome::Satisfied {
+                required_level: self.required_level,
+                observed_level: evidence.observed_level,
+            }
+        } else {
+            GreenEvidenceOutcome::Unsatisfied {
+                required_level: self.required_level,
+                observed_level: evidence.observed_level,
+                missing,
+                blocking_flakes,
+            }
+        }
+    }
+
+    #[must_use]
+    pub fn is_satisfied_by(&self, observed_level: GreenLevel) -> bool {
         observed_level >= self.required_level
     }
 }
 
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct GreenEvidence {
+    pub observed_level: GreenLevel,
+    pub test_commands: Vec<TestCommandProvenance>,
+    pub base_branch_fresh: bool,
+    pub known_flakes: Vec<KnownFlake>,
+    pub recovery_attempt_context_recorded: bool,
+}
+
+impl GreenEvidence {
+    #[must_use]
+    pub fn new(observed_level: GreenLevel) -> Self {
+        Self {
+            observed_level,
+            test_commands: Vec::new(),
+            base_branch_fresh: false,
+            known_flakes: Vec::new(),
+            recovery_attempt_context_recorded: false,
+        }
+    }
+
+    #[must_use]
+    pub fn with_test_command(mut self, command: impl Into<String>, exit_code: i32) -> Self {
+        self.test_commands.push(TestCommandProvenance {
+            command: command.into(),
+            exit_code,
+        });
+        self
+    }
+
+    #[must_use]
+    pub fn with_base_branch_fresh(mut self, is_fresh: bool) -> Self {
+        self.base_branch_fresh = is_fresh;
+        self
+    }
+
+    #[must_use]
+    pub fn with_known_flake(mut self, test_name: impl Into<String>, blocks_green: bool) -> Self {
+        self.known_flakes.push(KnownFlake {
+            test_name: test_name.into(),
+            blocks_green,
+        });
+        self
+    }
+
+    #[must_use]
+    pub fn with_recovery_attempt_context(mut self, recorded: bool) -> Self {
+        self.recovery_attempt_context_recorded = recorded;
+        self
+    }
+
+    #[must_use]
+    pub fn has_passing_test_command(&self) -> bool {
+        self.test_commands.iter().any(TestCommandProvenance::passed)
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct TestCommandProvenance {
+    pub command: String,
+    pub exit_code: i32,
+}
+
+impl TestCommandProvenance {
+    #[must_use]
+    pub fn passed(&self) -> bool {
+        self.exit_code == 0 && !self.command.trim().is_empty()
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct KnownFlake {
+    pub test_name: String,
+    pub blocks_green: bool,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum GreenContractRequirement {
+    RequiredLevel,
+    TestCommandProvenance,
+    BaseBranchFreshness,
+    RecoveryAttemptContext,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(tag = "outcome", rename_all = "snake_case")]
+pub enum GreenEvidenceOutcome {
+    Satisfied {
+        required_level: GreenLevel,
+        observed_level: GreenLevel,
+    },
+    Unsatisfied {
+        required_level: GreenLevel,
+        observed_level: GreenLevel,
+        missing: Vec<GreenContractRequirement>,
+        blocking_flakes: Vec<KnownFlake>,
+    },
+}
+
+impl GreenEvidenceOutcome {
+    #[must_use]
+    pub fn is_satisfied(&self) -> bool {
+        matches!(self, Self::Satisfied { .. })
+    }
+}
+
 #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
 #[serde(tag = "outcome", rename_all = "snake_case")]
 pub enum GreenContractOutcome {
@@ -149,4 +327,83 @@ mod tests {
             }
         );
     }
+    #[test]
+    fn merge_ready_contract_requires_provenance_beyond_test_level() {
+        // given
+        let contract = GreenContract::merge_ready(GreenLevel::Workspace);
+        let evidence = GreenEvidence::new(GreenLevel::Workspace)
+            .with_test_command("cargo test --manifest-path rust/Cargo.toml", 0);
+
+        // when
+        let outcome = contract.evaluate_evidence(&evidence);
+
+        // then
+        assert_eq!(
+            outcome,
+            GreenEvidenceOutcome::Unsatisfied {
+                required_level: GreenLevel::Workspace,
+                observed_level: GreenLevel::Workspace,
+                missing: vec![
+                    GreenContractRequirement::BaseBranchFreshness,
+                    GreenContractRequirement::RecoveryAttemptContext,
+                ],
+                blocking_flakes: vec![],
+            }
+        );
+        assert!(!outcome.is_satisfied());
+    }
+
+    #[test]
+    fn merge_ready_contract_accepts_complete_test_provenance_context() {
+        // given
+        let contract = GreenContract::merge_ready(GreenLevel::Workspace);
+        let evidence = GreenEvidence::new(GreenLevel::MergeReady)
+            .with_test_command("cargo test --manifest-path rust/Cargo.toml", 0)
+            .with_base_branch_fresh(true)
+            .with_recovery_attempt_context(true);
+
+        // when
+        let outcome = contract.evaluate_evidence(&evidence);
+
+        // then
+        assert_eq!(
+            outcome,
+            GreenEvidenceOutcome::Satisfied {
+                required_level: GreenLevel::Workspace,
+                observed_level: GreenLevel::MergeReady,
+            }
+        );
+    }
+
+    #[test]
+    fn known_blocking_flake_prevents_green_contract_satisfaction() {
+        // given
+        let contract = GreenContract::merge_ready(GreenLevel::Workspace);
+        let evidence = GreenEvidence::new(GreenLevel::MergeReady)
+            .with_test_command("cargo test --manifest-path rust/Cargo.toml", 0)
+            .with_base_branch_fresh(true)
+            .with_recovery_attempt_context(true)
+            .with_known_flake(
+                "session_lifecycle_prefers_running_process_over_idle_shell",
+                true,
+            );
+
+        // when
+        let outcome = contract.evaluate_evidence(&evidence);
+
+        // then
+        assert_eq!(
+            outcome,
+            GreenEvidenceOutcome::Unsatisfied {
+                required_level: GreenLevel::Workspace,
+                observed_level: GreenLevel::MergeReady,
+                missing: vec![],
+                blocking_flakes: vec![KnownFlake {
+                    test_name: "session_lifecycle_prefers_running_process_over_idle_shell"
+                        .to_string(),
+                    blocks_green: true,
+                }],
+            }
+        );
+    }
 }

rust/crates/runtime/src/lib.rs

@@ -4,6 +4,7 @@
 //! MCP plumbing, tool-facing file operations, and the core conversation loop
 //! that drives interactive and one-shot turns.
 
+mod approval_tokens;
 mod bash;
 pub mod bash_validation;
 mod bootstrap;
@@ -13,6 +14,7 @@ mod config;
 pub mod config_validate;
 mod conversation;
 mod file_ops;
+pub mod g004_conformance;
 mod git_context;
 pub mod green_contract;
 mod hooks;
@@ -33,6 +35,7 @@ mod policy_engine;
 mod prompt;
 pub mod recovery_recipes;
 mod remote;
+mod report_schema;
 pub mod sandbox;
 mod session;
 pub mod session_control;
@@ -49,6 +52,10 @@ mod trust_resolver;
 mod usage;
 pub mod worker_boot;
 
+pub use approval_tokens::{
+    ApprovalDelegationHop, ApprovalScope, ApprovalTokenAudit, ApprovalTokenError,
+    ApprovalTokenGrant, ApprovalTokenLedger, ApprovalTokenStatus,
+};
 pub use bash::{execute_bash, BashCommandInput, BashCommandOutput};
 pub use bootstrap::{BootstrapPhase, BootstrapPlan};
 pub use branch_lock::{detect_branch_lock_collisions, BranchLockCollision, BranchLockIntent};
@@ -74,9 +81,10 @@ pub use conversation::{
     ToolExecutor, TurnSummary,
 };
 pub use file_ops::{
-    edit_file, glob_search, grep_search, read_file, write_file, EditFileOutput, GlobSearchOutput,
+    edit_file, edit_file_in_workspace, glob_search, glob_search_in_workspace, grep_search,
-    GrepSearchInput, GrepSearchOutput, ReadFileOutput, StructuredPatchHunk, TextFilePayload,
+    grep_search_in_workspace, read_file, read_file_in_workspace, write_file,
-    WriteFileOutput,
+    write_file_in_workspace, EditFileOutput, GlobSearchOutput, GrepSearchInput, GrepSearchOutput,
+    ReadFileOutput, StructuredPatchHunk, TextFilePayload, WriteFileOutput,
 };
 pub use git_context::{GitCommitEntry, GitContext};
 pub use hooks::{
@@ -131,18 +139,26 @@ pub use policy_engine::{
     PolicyEngine, PolicyRule, ReconcileReason, ReviewStatus,
 };
 pub use prompt::{
-    load_system_prompt, prepend_bullets, ContextFile, ProjectContext, PromptBuildError,
+    load_system_prompt, prepend_bullets, ContextFile, ModelFamilyIdentity, ProjectContext,
-    SystemPromptBuilder, FRONTIER_MODEL_NAME, SYSTEM_PROMPT_DYNAMIC_BOUNDARY,
+    PromptBuildError, SystemPromptBuilder, FRONTIER_MODEL_NAME, SYSTEM_PROMPT_DYNAMIC_BOUNDARY,
 };
 pub use recovery_recipes::{
-    attempt_recovery, recipe_for, EscalationPolicy, FailureScenario, RecoveryContext,
+    attempt_recovery, recipe_for, EscalationPolicy, FailureScenario, RecoveryAttemptState,
-    RecoveryEvent, RecoveryRecipe, RecoveryResult, RecoveryStep,
+    RecoveryAttemptType, RecoveryCommandResult, RecoveryContext, RecoveryEvent,
+    RecoveryLedgerEntry, RecoveryRecipe, RecoveryResult, RecoveryStatusReport, RecoveryStep,
 };
 pub use remote::{
     inherited_upstream_proxy_env, no_proxy_list, read_token, upstream_proxy_ws_url,
     RemoteSessionContext, UpstreamProxyBootstrap, UpstreamProxyState, DEFAULT_REMOTE_BASE_URL,
     DEFAULT_SESSION_TOKEN_PATH, DEFAULT_SYSTEM_CA_BUNDLE, NO_PROXY_HOSTS, UPSTREAM_PROXY_ENV_KEYS,
 };
+pub use report_schema::{
+    canonicalize_report, project_report, report_content_hash, report_schema_v1_registry,
+    CanonicalReportV1, ClaimKind, ConsumerCapabilities, FieldDelta, FieldDeltaState,
+    NegativeEvidence, NegativeFindingStatus, ProjectionProvenance, RedactionProvenance,
+    ReportClaim, ReportConfidence, ReportIdentity, ReportProjectionV1, ReportSchemaField,
+    ReportSchemaRegistry, SensitivityClass, DEFAULT_PROJECTION_POLICY_V1, REPORT_SCHEMA_V1,
+};
 pub use sandbox::{
     build_linux_sandbox_command, detect_container_environment, detect_container_environment_from,
     resolve_sandbox_status, resolve_sandbox_status_for_request, ContainerEnvironment,

rust/crates/runtime/src/policy_engine.rs

@@ -2,7 +2,7 @@ use std::time::Duration;
 
 pub type GreenLevel = u8;
 
-const STALE_BRANCH_THRESHOLD: Duration = Duration::from_secs(60 * 60);
+const STALE_BRANCH_THRESHOLD: Duration = Duration::from_hours(1);
 
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub struct PolicyRule {
@@ -58,7 +58,9 @@ impl PolicyCondition {
             Self::Or(conditions) => conditions
                 .iter()
                 .any(|condition| condition.matches(context)),
-            Self::GreenAt { level } => context.green_level >= *level,
+            Self::GreenAt { level } => {
+                context.green_contract_satisfied && context.green_level >= *level
+            }
             Self::StaleBranch => context.branch_freshness >= STALE_BRANCH_THRESHOLD,
             Self::StartupBlocked => context.blocker == LaneBlocker::Startup,
             Self::LaneCompleted => context.completed,
@@ -134,6 +136,7 @@ pub enum DiffScope {
 pub struct LaneContext {
     pub lane_id: String,
     pub green_level: GreenLevel,
+    pub green_contract_satisfied: bool,
     pub branch_freshness: Duration,
     pub blocker: LaneBlocker,
     pub review_status: ReviewStatus,
@@ -156,6 +159,7 @@ impl LaneContext {
         Self {
             lane_id: lane_id.into(),
             green_level,
+            green_contract_satisfied: false,
             branch_freshness,
             blocker,
             review_status,
@@ -171,6 +175,7 @@ impl LaneContext {
         Self {
             lane_id: lane_id.into(),
             green_level: 0,
+            green_contract_satisfied: false,
             branch_freshness: Duration::from_secs(0),
             blocker: LaneBlocker::None,
             review_status: ReviewStatus::Pending,
@@ -179,6 +184,12 @@ impl LaneContext {
             reconciled: true,
         }
     }
+
+    #[must_use]
+    pub fn with_green_contract_satisfied(mut self, satisfied: bool) -> Self {
+        self.green_contract_satisfied = satisfied;
+        self
+    }
 }
 
 #[derive(Debug, Clone, PartialEq, Eq)]
@@ -257,7 +268,8 @@ mod tests {
             ReviewStatus::Approved,
             DiffScope::Scoped,
             false,
-        );
+        )
+        .with_green_contract_satisfied(true);
 
         // when
         let actions = engine.evaluate(&context);
@@ -266,6 +278,36 @@ mod tests {
         assert_eq!(actions, vec![PolicyAction::MergeToDev]);
     }
 
+    #[test]
+    fn merge_rule_blocks_when_green_tests_lack_contract_provenance() {
+        // given
+        let engine = PolicyEngine::new(vec![PolicyRule::new(
+            "merge-to-dev",
+            PolicyCondition::And(vec![
+                PolicyCondition::GreenAt { level: 2 },
+                PolicyCondition::ScopedDiff,
+                PolicyCondition::ReviewPassed,
+            ]),
+            PolicyAction::MergeToDev,
+            20,
+        )]);
+        let context = LaneContext::new(
+            "lane-7",
+            3,
+            Duration::from_secs(5),
+            LaneBlocker::None,
+            ReviewStatus::Approved,
+            DiffScope::Scoped,
+            false,
+        );
+
+        // when
+        let actions = engine.evaluate(&context);
+
+        // then
+        assert!(actions.is_empty());
+    }
+
     #[test]
     fn stale_branch_rule_fires_at_threshold() {
         // given
@@ -468,7 +510,8 @@ mod tests {
             ReviewStatus::Pending,
             DiffScope::Full,
             false,
-        );
+        )
+        .with_green_contract_satisfied(true);
 
         // when
         let actions = engine.evaluate(&context);

rust/crates/runtime/src/prompt.rs

@@ -43,6 +43,24 @@ pub const FRONTIER_MODEL_NAME: &str = "Claude Opus 4.6";
 const MAX_INSTRUCTION_FILE_CHARS: usize = 4_000;
 const MAX_TOTAL_INSTRUCTION_CHARS: usize = 12_000;
 
+/// Neutral identity for the model family line in generated prompts.
+#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
+pub enum ModelFamilyIdentity {
+    #[default]
+    Claude,
+    Generic,
+}
+
+impl ModelFamilyIdentity {
+    #[must_use]
+    pub const fn family_label(self) -> &'static str {
+        match self {
+            Self::Claude => FRONTIER_MODEL_NAME,
+            Self::Generic => "an AI assistant",
+        }
+    }
+}
+
 /// Contents of an instruction file included in prompt construction.
 #[derive(Debug, Clone, PartialEq, Eq)]
 pub struct ContextFile {
@@ -97,6 +115,7 @@ pub struct SystemPromptBuilder {
     output_style_prompt: Option<String>,
     os_name: Option<String>,
     os_version: Option<String>,
+    model_family: Option<ModelFamilyIdentity>,
     append_sections: Vec<String>,
     project_context: Option<ProjectContext>,
     config: Option<RuntimeConfig>,
@@ -122,6 +141,12 @@ impl SystemPromptBuilder {
         self
     }
 
+    #[must_use]
+    pub fn with_model_family(mut self, model_family: ModelFamilyIdentity) -> Self {
+        self.model_family = Some(model_family);
+        self
+    }
+
     #[must_use]
     pub fn with_project_context(mut self, project_context: ProjectContext) -> Self {
         self.project_context = Some(project_context);
@@ -179,9 +204,10 @@ impl SystemPromptBuilder {
             || "unknown".to_string(),
             |context| context.current_date.clone(),
         );
+        let identity = self.model_family.unwrap_or_default();
         let mut lines = vec!["# Environment context".to_string()];
         lines.extend(prepend_bullets(vec![
-            format!("Model family: {FRONTIER_MODEL_NAME}"),
+            format!("Model family: {}", identity.family_label()),
             format!("Working directory: {cwd}"),
             format!("Date: {date}"),
             format!(
@@ -434,12 +460,14 @@ pub fn load_system_prompt(
     current_date: impl Into<String>,
     os_name: impl Into<String>,
     os_version: impl Into<String>,
+    model_family: ModelFamilyIdentity,
 ) -> Result<Vec<String>, PromptBuildError> {
     let cwd = cwd.into();
     let project_context = ProjectContext::discover_with_git(&cwd, current_date.into())?;
     let config = ConfigLoader::default_for(&cwd).load()?;
     Ok(SystemPromptBuilder::new()
         .with_os(os_name, os_version)
+        .with_model_family(model_family)
         .with_project_context(project_context)
         .with_runtime_config(config)
         .build())
@@ -522,7 +550,8 @@ mod tests {
     use super::{
         collapse_blank_lines, display_context_path, normalize_instruction_content,
         render_instruction_content, render_instruction_files, truncate_instruction_content,
-        ContextFile, ProjectContext, SystemPromptBuilder, SYSTEM_PROMPT_DYNAMIC_BOUNDARY,
+        ContextFile, ModelFamilyIdentity, ProjectContext, SystemPromptBuilder,
+        SYSTEM_PROMPT_DYNAMIC_BOUNDARY,
     };
     use crate::config::ConfigLoader;
     use std::fs;
@@ -804,13 +833,19 @@ mod tests {
         std::env::set_var("HOME", &root);
         std::env::set_var("CLAW_CONFIG_HOME", root.join("missing-home"));
         std::env::set_current_dir(&root).expect("change cwd");
-        let prompt = super::load_system_prompt(&root, "2026-03-31", "linux", "6.8")
+        let prompt = super::load_system_prompt(
-            .expect("system prompt should load")
+            &root,
-            .join(
+            "2026-03-31",
-                "
+            "linux",
+            "6.8",
+            ModelFamilyIdentity::Claude,
+        )
+        .expect("system prompt should load")
+        .join(
+            "
 
 ",
-            );
+        );
         std::env::set_current_dir(previous).expect("restore cwd");
         if let Some(value) = original_home {
             std::env::set_var("HOME", value);
@@ -828,6 +863,50 @@ mod tests {
         fs::remove_dir_all(root).expect("cleanup temp dir");
     }
 
+    #[test]
+    fn renders_default_claude_model_family_identity() {
+        // given: a prompt builder without an explicit model family override
+        let project_context = ProjectContext {
+            cwd: PathBuf::from("/tmp/project"),
+            current_date: "2026-03-31".to_string(),
+            ..ProjectContext::default()
+        };
+
+        // when: rendering the system prompt environment section
+        let prompt = SystemPromptBuilder::new()
+            .with_os("linux", "6.8")
+            .with_project_context(project_context)
+            .render();
+
+        // then: the Claude model family label is preserved by default
+        assert!(prompt.contains("Model family: Claude Opus 4.6"));
+    }
+
+    #[test]
+    fn renders_generic_model_family_identity_without_claude_label() {
+        // given: a prompt builder with generic model family identity
+        let project_context = ProjectContext {
+            cwd: PathBuf::from("/tmp/project"),
+            current_date: "2026-03-31".to_string(),
+            ..ProjectContext::default()
+        };
+
+        // when: rendering the system prompt environment section
+        let prompt = SystemPromptBuilder::new()
+            .with_os("linux", "6.8")
+            .with_model_family(ModelFamilyIdentity::Generic)
+            .with_project_context(project_context)
+            .render();
+        let model_family_line = prompt
+            .lines()
+            .find(|line| line.contains("Model family:"))
+            .expect("model family line should render");
+
+        // then: the model family line is neutral and excludes Claude Opus 4.6
+        assert_eq!(model_family_line, " - Model family: an AI assistant");
+        assert!(!model_family_line.contains("Claude Opus 4.6"));
+    }
+
     #[test]
     fn renders_claude_code_style_sections_with_project_context() {
         let root = temp_dir();

rust/crates/runtime/src/recovery_recipes.rs

@@ -45,7 +45,9 @@ impl FailureScenario {
     #[must_use]
     pub fn from_worker_failure_kind(kind: WorkerFailureKind) -> Self {
         match kind {
-            WorkerFailureKind::TrustGate => Self::TrustPromptUnresolved,
+            WorkerFailureKind::TrustGate | WorkerFailureKind::ToolPermissionGate => {
+                Self::TrustPromptUnresolved
+            }
             WorkerFailureKind::PromptDelivery => Self::PromptMisdelivery,
             WorkerFailureKind::Protocol => Self::McpHandshakeFailure,
             WorkerFailureKind::Provider | WorkerFailureKind::StartupNoEvidence => {
@@ -119,6 +121,21 @@ pub enum RecoveryResult {
     },
 }
 
+/// Type of recovery execution represented in the ledger.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum RecoveryAttemptType {
+    Automatic,
+}
+
+/// Result for one executable recovery command/step.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct RecoveryCommandResult {
+    pub command: RecoveryStep,
+    pub status: RecoveryAttemptState,
+    pub result: String,
+}
+
 /// Structured event emitted during recovery.
 #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
 #[serde(rename_all = "snake_case")]
@@ -133,14 +150,59 @@ pub enum RecoveryEvent {
     Escalated,
 }
 
+/// Machine-readable recovery progress for one failure scenario.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct RecoveryLedgerEntry {
+    pub recipe_id: String,
+    pub attempt_type: RecoveryAttemptType,
+    pub trigger: FailureScenario,
+    pub attempt_count: u32,
+    pub retry_limit: u32,
+    pub attempts_remaining: u32,
+    pub state: RecoveryAttemptState,
+    pub started_at: Option<String>,
+    pub finished_at: Option<String>,
+    pub command_results: Vec<RecoveryCommandResult>,
+    pub result: Option<RecoveryResult>,
+    pub last_failure_summary: Option<String>,
+    pub escalation_reason: Option<String>,
+}
+
+/// Current state of a recovery recipe attempt.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum RecoveryAttemptState {
+    Queued,
+    Running,
+    Succeeded,
+    Failed,
+    Exhausted,
+}
+
+/// Machine-readable status projection for callers that need to
+/// distinguish an untouched scenario from an exhausted recovery.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct RecoveryStatusReport {
+    pub scenario: FailureScenario,
+    pub attempted: bool,
+    pub state: Option<RecoveryAttemptState>,
+    pub attempt_count: u32,
+    pub retry_limit: Option<u32>,
+    pub attempts_remaining: Option<u32>,
+    pub escalation_reason: Option<String>,
+}
+
 /// Minimal context for tracking recovery state and emitting events.
 ///
-/// Holds per-scenario attempt counts, a structured event log, and an
+/// Holds per-scenario attempt counts, a structured event log, a recovery
-/// optional simulation knob for controlling step outcomes during tests.
+/// attempt ledger, and an optional simulation knob for controlling step
+/// outcomes during tests.
 #[derive(Debug, Clone, Default)]
 pub struct RecoveryContext {
     attempts: HashMap<FailureScenario, u32>,
     events: Vec<RecoveryEvent>,
+    ledger: HashMap<FailureScenario, RecoveryLedgerEntry>,
+    clock_tick: u64,
     /// Optional step index at which simulated execution fails.
     /// `None` means all steps succeed.
     fail_at_step: Option<usize>,
@@ -170,6 +232,51 @@ impl RecoveryContext {
     pub fn attempt_count(&self, scenario: &FailureScenario) -> u32 {
         self.attempts.get(scenario).copied().unwrap_or(0)
     }
+
+    /// Returns the machine-readable recovery ledger entry for a scenario.
+    #[must_use]
+    pub fn ledger_entry(&self, scenario: &FailureScenario) -> Option<&RecoveryLedgerEntry> {
+        self.ledger.get(scenario)
+    }
+
+    /// Returns all recovery ledger entries currently tracked by this context.
+    #[must_use]
+    pub fn ledger_entries(&self) -> Vec<&RecoveryLedgerEntry> {
+        let mut entries: Vec<_> = self.ledger.values().collect();
+        entries.sort_by(|left, right| left.recipe_id.cmp(&right.recipe_id));
+        entries
+    }
+
+    /// Returns a compact machine-readable recovery status for a scenario,
+    /// including `attempted = false` when no ledger entry exists yet.
+    #[must_use]
+    pub fn status_report(&self, scenario: &FailureScenario) -> RecoveryStatusReport {
+        self.ledger_entry(scenario).map_or(
+            RecoveryStatusReport {
+                scenario: *scenario,
+                attempted: false,
+                state: None,
+                attempt_count: 0,
+                retry_limit: None,
+                attempts_remaining: None,
+                escalation_reason: None,
+            },
+            |entry| RecoveryStatusReport {
+                scenario: *scenario,
+                attempted: entry.attempt_count > 0,
+                state: Some(entry.state),
+                attempt_count: entry.attempt_count,
+                retry_limit: Some(entry.retry_limit),
+                attempts_remaining: Some(entry.attempts_remaining),
+                escalation_reason: entry.escalation_reason.clone(),
+            },
+        )
+    }
+
+    fn next_timestamp(&mut self) -> String {
+        self.clock_tick += 1;
+        format!("recovery-ledger-tick-{}", self.clock_tick)
+    }
 }
 
 /// Returns the known recovery recipe for the given failure scenario.
@@ -231,18 +338,51 @@ pub fn recipe_for(scenario: &FailureScenario) -> RecoveryRecipe {
 /// Looks up the recipe, enforces the one-attempt-before-escalation
 /// policy, simulates step execution (controlled by the context), and
 /// emits structured [`RecoveryEvent`]s for every attempt.
+#[allow(clippy::too_many_lines)]
 pub fn attempt_recovery(scenario: &FailureScenario, ctx: &mut RecoveryContext) -> RecoveryResult {
     let recipe = recipe_for(scenario);
-    let attempt_count = ctx.attempts.entry(*scenario).or_insert(0);
+    let recipe_id = scenario.to_string();
+    ctx.ledger
+        .entry(*scenario)
+        .or_insert_with(|| RecoveryLedgerEntry {
+            recipe_id: recipe_id.clone(),
+            attempt_type: RecoveryAttemptType::Automatic,
+            trigger: *scenario,
+            attempt_count: 0,
+            retry_limit: recipe.max_attempts,
+            attempts_remaining: recipe.max_attempts,
+            state: RecoveryAttemptState::Queued,
+            started_at: None,
+            finished_at: None,
+            command_results: Vec::new(),
+            result: None,
+            last_failure_summary: None,
+            escalation_reason: None,
+        });
+
+    let current_attempts = ctx.attempt_count(scenario);
 
     // Enforce one automatic recovery attempt before escalation.
-    if *attempt_count >= recipe.max_attempts {
+    if current_attempts >= recipe.max_attempts {
         let result = RecoveryResult::EscalationRequired {
             reason: format!(
                 "max recovery attempts ({}) exceeded for {}",
                 recipe.max_attempts, scenario
             ),
         };
+        let finished_at = ctx.next_timestamp();
+        if let Some(entry) = ctx.ledger.get_mut(scenario) {
+            entry.attempt_count = current_attempts;
+            entry.attempts_remaining = 0;
+            entry.state = RecoveryAttemptState::Exhausted;
+            entry.finished_at = Some(finished_at);
+            entry.result = Some(result.clone());
+            let RecoveryResult::EscalationRequired { reason } = &result else {
+                unreachable!("exhaustion always produces escalation");
+            };
+            entry.last_failure_summary = Some(reason.clone());
+            entry.escalation_reason = Some(reason.clone());
+        }
         ctx.events.push(RecoveryEvent::RecoveryAttempted {
             scenario: *scenario,
             recipe,
@@ -252,19 +392,44 @@ pub fn attempt_recovery(scenario: &FailureScenario, ctx: &mut RecoveryContext) -
         return result;
     }
 
-    *attempt_count += 1;
+    let updated_attempts = ctx.attempts.entry(*scenario).or_insert(0);
+    *updated_attempts += 1;
+    let updated_attempts = *updated_attempts;
+    let started_at = ctx.next_timestamp();
+    if let Some(entry) = ctx.ledger.get_mut(scenario) {
+        entry.attempt_count = updated_attempts;
+        entry.attempts_remaining = recipe.max_attempts.saturating_sub(updated_attempts);
+        entry.state = RecoveryAttemptState::Running;
+        entry.started_at = Some(started_at);
+        entry.finished_at = None;
+        entry.command_results.clear();
+        entry.result = None;
+        entry.last_failure_summary = None;
+        entry.escalation_reason = None;
+    }
 
     // Execute steps, honoring the optional fail_at_step simulation.
     let fail_index = ctx.fail_at_step;
     let mut executed = Vec::new();
+    let mut command_results = Vec::new();
     let mut failed = false;
 
     for (i, step) in recipe.steps.iter().enumerate() {
         if fail_index == Some(i) {
+            command_results.push(RecoveryCommandResult {
+                command: step.clone(),
+                status: RecoveryAttemptState::Failed,
+                result: format!("step {i} failed for {scenario}"),
+            });
             failed = true;
             break;
         }
         executed.push(step.clone());
+        command_results.push(RecoveryCommandResult {
+            command: step.clone(),
+            status: RecoveryAttemptState::Succeeded,
+            result: format!("step {i} succeeded for {scenario}"),
+        });
     }
 
     let result = if failed {
@@ -286,6 +451,29 @@ pub fn attempt_recovery(scenario: &FailureScenario, ctx: &mut RecoveryContext) -
     };
 
     // Emit the attempt as structured event data.
+    let finished_at = ctx.next_timestamp();
+    if let Some(entry) = ctx.ledger.get_mut(scenario) {
+        entry.finished_at = Some(finished_at);
+        entry.command_results = command_results;
+        entry.result = Some(result.clone());
+        match &result {
+            RecoveryResult::Recovered { .. } => {
+                entry.state = RecoveryAttemptState::Succeeded;
+            }
+            RecoveryResult::PartialRecovery { remaining, .. } => {
+                entry.state = RecoveryAttemptState::Failed;
+                entry.last_failure_summary = Some(format!(
+                    "{} step(s) remaining after partial recovery",
+                    remaining.len()
+                ));
+            }
+            RecoveryResult::EscalationRequired { reason } => {
+                entry.state = RecoveryAttemptState::Exhausted;
+                entry.last_failure_summary = Some(reason.clone());
+                entry.escalation_reason = Some(reason.clone());
+            }
+        }
+    }
     ctx.events.push(RecoveryEvent::RecoveryAttempted {
         scenario: *scenario,
         recipe,
@@ -497,6 +685,126 @@ mod tests {
         assert_eq!(ctx.attempt_count(&FailureScenario::PromptMisdelivery), 0);
     }
 
+    #[test]
+    fn recovery_context_exposes_machine_readable_ledger() {
+        // given
+        let mut ctx = RecoveryContext::new();
+
+        // when
+        let result = attempt_recovery(&FailureScenario::StaleBranch, &mut ctx);
+
+        // then
+        assert_eq!(result, RecoveryResult::Recovered { steps_taken: 2 });
+        let entry = ctx
+            .ledger_entry(&FailureScenario::StaleBranch)
+            .expect("stale branch ledger entry");
+        assert_eq!(entry.recipe_id, "stale_branch");
+        assert_eq!(entry.attempt_type, RecoveryAttemptType::Automatic);
+        assert_eq!(entry.trigger, FailureScenario::StaleBranch);
+        assert_eq!(entry.attempt_count, 1);
+        assert_eq!(entry.retry_limit, 1);
+        assert_eq!(entry.attempts_remaining, 0);
+        assert_eq!(entry.state, RecoveryAttemptState::Succeeded);
+        assert!(entry.started_at.is_some());
+        assert!(entry.finished_at.is_some());
+        assert_eq!(
+            entry.result,
+            Some(RecoveryResult::Recovered { steps_taken: 2 })
+        );
+        assert_eq!(entry.command_results.len(), 2);
+        assert_eq!(entry.command_results[0].command, RecoveryStep::RebaseBranch);
+        assert_eq!(
+            entry.command_results[0].status,
+            RecoveryAttemptState::Succeeded
+        );
+        assert_eq!(entry.last_failure_summary, None);
+        assert_eq!(entry.escalation_reason, None);
+    }
+
+    #[test]
+    fn recovery_ledger_records_exhausted_escalation_reason() {
+        // given
+        let mut ctx = RecoveryContext::new();
+        let scenario = FailureScenario::PromptMisdelivery;
+
+        // when
+        let _ = attempt_recovery(&scenario, &mut ctx);
+        let result = attempt_recovery(&scenario, &mut ctx);
+
+        // then
+        assert!(matches!(result, RecoveryResult::EscalationRequired { .. }));
+        let entry = ctx.ledger_entry(&scenario).expect("ledger entry");
+        assert_eq!(entry.state, RecoveryAttemptState::Exhausted);
+        assert_eq!(entry.attempt_count, 1);
+        assert_eq!(entry.attempts_remaining, 0);
+        assert!(matches!(
+            entry.result,
+            Some(RecoveryResult::EscalationRequired { .. })
+        ));
+        assert!(entry
+            .escalation_reason
+            .as_deref()
+            .expect("escalation reason")
+            .contains("max recovery attempts"));
+    }
+
+    #[test]
+    fn recovery_status_report_distinguishes_not_attempted_from_exhausted() {
+        // given
+        let mut ctx = RecoveryContext::new();
+        let scenario = FailureScenario::PromptMisdelivery;
+
+        // then — no ledger entry is not the same as exhausted.
+        let not_attempted = ctx.status_report(&scenario);
+        assert!(!not_attempted.attempted);
+        assert_eq!(not_attempted.state, None);
+        assert_eq!(not_attempted.attempt_count, 0);
+        assert_eq!(not_attempted.retry_limit, None);
+
+        // when — one allowed attempt then one extra attempt.
+        let _ = attempt_recovery(&scenario, &mut ctx);
+        let _ = attempt_recovery(&scenario, &mut ctx);
+
+        // then
+        let exhausted = ctx.status_report(&scenario);
+        assert!(exhausted.attempted);
+        assert_eq!(exhausted.state, Some(RecoveryAttemptState::Exhausted));
+        assert_eq!(exhausted.attempt_count, 1);
+        assert_eq!(exhausted.retry_limit, Some(1));
+        assert_eq!(exhausted.attempts_remaining, Some(0));
+        assert!(exhausted
+            .escalation_reason
+            .as_deref()
+            .is_some_and(|reason| reason.contains("max recovery attempts")));
+    }
+
+    #[test]
+    fn recovery_ledger_records_failed_command_result() {
+        // given
+        let mut ctx = RecoveryContext::new().with_fail_at_step(1);
+        let scenario = FailureScenario::PartialPluginStartup;
+
+        // when
+        let result = attempt_recovery(&scenario, &mut ctx);
+
+        // then
+        assert!(matches!(result, RecoveryResult::PartialRecovery { .. }));
+        let entry = ctx.ledger_entry(&scenario).expect("ledger entry");
+        assert_eq!(entry.state, RecoveryAttemptState::Failed);
+        assert_eq!(entry.command_results.len(), 2);
+        assert_eq!(
+            entry.command_results[0].status,
+            RecoveryAttemptState::Succeeded
+        );
+        assert_eq!(
+            entry.command_results[1].status,
+            RecoveryAttemptState::Failed
+        );
+        assert!(entry.command_results[1]
+            .result
+            .contains("partial_plugin_startup"));
+    }
+
     #[test]
     fn stale_branch_recipe_has_rebase_then_clean_build() {
         // given

rust/crates/runtime/src/report_schema.rs

@@ -0,0 +1,552 @@
+use std::collections::{BTreeMap, BTreeSet};
+
+use serde::{Deserialize, Serialize};
+use serde_json::Value;
+use sha2::{Digest, Sha256};
+
+pub const REPORT_SCHEMA_V1: &str = "claw.report.v1";
+pub const DEFAULT_PROJECTION_POLICY_V1: &str = "claw.report.projection.v1";
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum ClaimKind {
+    ObservedFact,
+    Inference,
+    Hypothesis,
+    Recommendation,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum ReportConfidence {
+    High,
+    Medium,
+    Low,
+    Unknown,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum SensitivityClass {
+    Public,
+    Internal,
+    OperatorOnly,
+    Secret,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum FieldDeltaState {
+    Changed,
+    Unchanged,
+    Cleared,
+    CarriedForward,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum NegativeFindingStatus {
+    NotObservedInCheckedScope,
+    UnknownNotChecked,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct ReportClaim {
+    pub id: String,
+    pub kind: ClaimKind,
+    pub text: String,
+    pub confidence: ReportConfidence,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub evidence: Vec<String>,
+    pub sensitivity: SensitivityClass,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct NegativeEvidence {
+    pub id: String,
+    pub status: NegativeFindingStatus,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub checked_surfaces: Vec<String>,
+    pub query: String,
+    pub window: String,
+    pub sensitivity: SensitivityClass,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct FieldDelta {
+    pub field: String,
+    pub state: FieldDeltaState,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub previous_hash: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub current_hash: Option<String>,
+    pub attribution: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct ReportIdentity {
+    pub report_id: String,
+    pub content_hash: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct CanonicalReportV1 {
+    pub schema_version: String,
+    pub identity: ReportIdentity,
+    pub generated_at: String,
+    pub producer: String,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub claims: Vec<ReportClaim>,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub negative_evidence: Vec<NegativeEvidence>,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub field_deltas: Vec<FieldDelta>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct ConsumerCapabilities {
+    pub consumer: String,
+    #[serde(default, skip_serializing_if = "BTreeSet::is_empty")]
+    pub schema_versions: BTreeSet<String>,
+    #[serde(default, skip_serializing_if = "BTreeSet::is_empty")]
+    pub field_families: BTreeSet<String>,
+    pub max_sensitivity: SensitivityClass,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct RedactionProvenance {
+    pub field_path: String,
+    pub reason: String,
+    pub policy_id: String,
+    pub original_hash: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct ProjectionProvenance {
+    pub policy_id: String,
+    pub source_schema_version: String,
+    pub source_report_id: String,
+    pub source_content_hash: String,
+    pub consumer: String,
+    pub downgraded: bool,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub omitted_field_families: Vec<String>,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub redactions: Vec<RedactionProvenance>,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct ReportProjectionV1 {
+    pub schema_version: String,
+    pub projection_id: String,
+    pub view: String,
+    pub provenance: ProjectionProvenance,
+    pub payload: Value,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct ReportSchemaField {
+    pub id: String,
+    pub description: String,
+    pub required: bool,
+    pub field_family: String,
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct ReportSchemaRegistry {
+    pub schema_version: String,
+    pub compatibility: String,
+    pub fields: Vec<ReportSchemaField>,
+}
+
+#[must_use]
+pub fn report_schema_v1_registry() -> ReportSchemaRegistry {
+    ReportSchemaRegistry {
+        schema_version: REPORT_SCHEMA_V1.to_string(),
+        compatibility: "additive fields are compatible; missing required fields are breaking"
+            .to_string(),
+        fields: vec![
+            field(
+                "identity.report_id",
+                "stable canonical report identity",
+                true,
+                "identity",
+            ),
+            field(
+                "identity.content_hash",
+                "hash of canonical payload excluding identity",
+                true,
+                "identity",
+            ),
+            field(
+                "claims[].kind",
+                "fact/inference/hypothesis/recommendation label",
+                true,
+                "claims",
+            ),
+            field(
+                "claims[].confidence",
+                "confidence bucket for the claim",
+                true,
+                "claims",
+            ),
+            field(
+                "claims[].evidence",
+                "evidence ids supporting a claim",
+                false,
+                "claims",
+            ),
+            field(
+                "negative_evidence[]",
+                "searched-and-not-found findings with checked scope",
+                false,
+                "negative_evidence",
+            ),
+            field(
+                "field_deltas[]",
+                "field-level changed/unchanged/cleared/carried-forward attribution",
+                false,
+                "field_deltas",
+            ),
+            field(
+                "projection.provenance.redactions[]",
+                "redaction policy provenance for projected fields",
+                false,
+                "projection",
+            ),
+        ],
+    }
+}
+
+#[must_use]
+pub fn canonicalize_report(mut report: CanonicalReportV1) -> CanonicalReportV1 {
+    report.schema_version = REPORT_SCHEMA_V1.to_string();
+    report.claims.sort_by(|a, b| a.id.cmp(&b.id));
+    report.negative_evidence.sort_by(|a, b| a.id.cmp(&b.id));
+    report.field_deltas.sort_by(|a, b| a.field.cmp(&b.field));
+    let content_hash = report_content_hash(&report);
+    if report.identity.report_id.is_empty() {
+        report.identity.report_id = format!("report-{content_hash}");
+    }
+    report.identity.content_hash = content_hash;
+    report
+}
+
+#[must_use]
+pub fn report_content_hash(report: &CanonicalReportV1) -> String {
+    let mut hashable = report.clone();
+    hashable.identity.report_id.clear();
+    hashable.identity.content_hash.clear();
+    stable_json_hash(&serde_json::to_value(hashable).expect("report should serialize"))
+}
+
+#[must_use]
+pub fn project_report(
+    report: &CanonicalReportV1,
+    capabilities: &ConsumerCapabilities,
+    view: impl Into<String>,
+) -> ReportProjectionV1 {
+    let view = view.into();
+    let supports_schema = capabilities.schema_versions.contains(REPORT_SCHEMA_V1);
+    let mut omitted_field_families = Vec::new();
+    let mut redactions = Vec::new();
+    let mut payload = serde_json::Map::new();
+
+    payload.insert(
+        "identity".to_string(),
+        serde_json::to_value(&report.identity).expect("identity serializes"),
+    );
+    payload.insert(
+        "generated_at".to_string(),
+        Value::String(report.generated_at.clone()),
+    );
+    payload.insert(
+        "producer".to_string(),
+        Value::String(report.producer.clone()),
+    );
+
+    if supports_family(capabilities, "claims") {
+        let claims = report
+            .claims
+            .iter()
+            .enumerate()
+            .filter_map(|(index, claim)| redact_claim(index, claim, capabilities, &mut redactions))
+            .collect::<Vec<_>>();
+        payload.insert("claims".to_string(), Value::Array(claims));
+    } else {
+        omitted_field_families.push("claims".to_string());
+    }
+
+    if supports_family(capabilities, "negative_evidence") {
+        payload.insert(
+            "negative_evidence".to_string(),
+            serde_json::to_value(&report.negative_evidence).expect("negative evidence serializes"),
+        );
+    } else {
+        omitted_field_families.push("negative_evidence".to_string());
+    }
+
+    if supports_family(capabilities, "field_deltas") {
+        payload.insert(
+            "field_deltas".to_string(),
+            serde_json::to_value(&report.field_deltas).expect("field deltas serialize"),
+        );
+    } else {
+        omitted_field_families.push("field_deltas".to_string());
+    }
+
+    let downgraded =
+        !supports_schema || !omitted_field_families.is_empty() || !redactions.is_empty();
+    let provenance = ProjectionProvenance {
+        policy_id: DEFAULT_PROJECTION_POLICY_V1.to_string(),
+        source_schema_version: report.schema_version.clone(),
+        source_report_id: report.identity.report_id.clone(),
+        source_content_hash: report.identity.content_hash.clone(),
+        consumer: capabilities.consumer.clone(),
+        downgraded,
+        omitted_field_families,
+        redactions,
+    };
+    let mut projection = ReportProjectionV1 {
+        schema_version: REPORT_SCHEMA_V1.to_string(),
+        projection_id: String::new(),
+        view,
+        provenance,
+        payload: Value::Object(payload),
+    };
+    projection.projection_id = stable_json_hash(&serde_json::json!({
+        "view": projection.view,
+        "provenance": projection.provenance,
+        "payload": projection.payload,
+    }));
+    projection
+}
+
+fn field(id: &str, description: &str, required: bool, field_family: &str) -> ReportSchemaField {
+    ReportSchemaField {
+        id: id.to_string(),
+        description: description.to_string(),
+        required,
+        field_family: field_family.to_string(),
+    }
+}
+
+fn supports_family(capabilities: &ConsumerCapabilities, family: &str) -> bool {
+    capabilities.field_families.is_empty() || capabilities.field_families.contains(family)
+}
+
+fn redact_claim(
+    index: usize,
+    claim: &ReportClaim,
+    capabilities: &ConsumerCapabilities,
+    redactions: &mut Vec<RedactionProvenance>,
+) -> Option<Value> {
+    if claim.sensitivity <= capabilities.max_sensitivity {
+        return Some(serde_json::to_value(claim).expect("claim serializes"));
+    }
+    if claim.sensitivity == SensitivityClass::Secret {
+        redactions.push(RedactionProvenance {
+            field_path: format!("claims[{index}]"),
+            reason: "omitted: sensitivity exceeds consumer policy".to_string(),
+            policy_id: DEFAULT_PROJECTION_POLICY_V1.to_string(),
+            original_hash: stable_json_hash(
+                &serde_json::to_value(claim).expect("claim serializes"),
+            ),
+        });
+        return None;
+    }
+
+    let mut redacted = claim.clone();
+    let original_hash = stable_json_hash(&serde_json::to_value(claim).expect("claim serializes"));
+    redacted.text = "<redacted>".to_string();
+    redacted.evidence.clear();
+    redactions.push(RedactionProvenance {
+        field_path: format!("claims[{index}].text"),
+        reason: "transformed: sensitivity exceeds consumer policy".to_string(),
+        policy_id: DEFAULT_PROJECTION_POLICY_V1.to_string(),
+        original_hash,
+    });
+    Some(serde_json::to_value(redacted).expect("redacted claim serializes"))
+}
+
+fn stable_json_hash(value: &Value) -> String {
+    let normalized = normalize_json(value);
+    let bytes = serde_json::to_vec(&normalized).expect("normalized json should serialize");
+    let digest = Sha256::digest(bytes);
+    let mut hash = String::with_capacity(16);
+    for byte in &digest[..8] {
+        use std::fmt::Write as _;
+        write!(&mut hash, "{byte:02x}").expect("writing to String should not fail");
+    }
+    hash
+}
+
+fn normalize_json(value: &Value) -> Value {
+    match value {
+        Value::Array(values) => Value::Array(values.iter().map(normalize_json).collect()),
+        Value::Object(map) => {
+            let sorted = map
+                .iter()
+                .map(|(key, value)| (key.clone(), normalize_json(value)))
+                .collect::<BTreeMap<_, _>>();
+            serde_json::to_value(sorted).expect("sorted map should serialize")
+        }
+        other => other.clone(),
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{
+        canonicalize_report, project_report, report_schema_v1_registry, CanonicalReportV1,
+        ClaimKind, ConsumerCapabilities, FieldDelta, FieldDeltaState, NegativeEvidence,
+        NegativeFindingStatus, ReportClaim, ReportConfidence, ReportIdentity, SensitivityClass,
+        REPORT_SCHEMA_V1,
+    };
+
+    fn fixture_report() -> CanonicalReportV1 {
+        canonicalize_report(CanonicalReportV1 {
+            schema_version: String::new(),
+            identity: ReportIdentity {
+                report_id: String::new(),
+                content_hash: String::new(),
+            },
+            generated_at: "2026-05-14T00:00:00Z".to_string(),
+            producer: "worker-1".to_string(),
+            claims: vec![
+                ReportClaim {
+                    id: "claim-secret".to_string(),
+                    kind: ClaimKind::ObservedFact,
+                    text: "secret token appeared in logs".to_string(),
+                    confidence: ReportConfidence::High,
+                    evidence: vec!["log:secret".to_string()],
+                    sensitivity: SensitivityClass::Secret,
+                },
+                ReportClaim {
+                    id: "claim-hypothesis".to_string(),
+                    kind: ClaimKind::Hypothesis,
+                    text: "transport restart likely caused the retry".to_string(),
+                    confidence: ReportConfidence::Medium,
+                    evidence: vec!["event:transport".to_string()],
+                    sensitivity: SensitivityClass::Internal,
+                },
+                ReportClaim {
+                    id: "claim-fact".to_string(),
+                    kind: ClaimKind::ObservedFact,
+                    text: "lane finished once".to_string(),
+                    confidence: ReportConfidence::High,
+                    evidence: vec!["event:lane.finished".to_string()],
+                    sensitivity: SensitivityClass::Public,
+                },
+            ],
+            negative_evidence: vec![NegativeEvidence {
+                id: "neg-blocker".to_string(),
+                status: NegativeFindingStatus::NotObservedInCheckedScope,
+                checked_surfaces: vec!["lane_events".to_string(), "worker_status".to_string()],
+                query: "current blocker".to_string(),
+                window: "2026-05-14T00:00:00Z/2026-05-14T00:05:00Z".to_string(),
+                sensitivity: SensitivityClass::Public,
+            }],
+            field_deltas: vec![FieldDelta {
+                field: "blocker".to_string(),
+                state: FieldDeltaState::Cleared,
+                previous_hash: Some("prev123".to_string()),
+                current_hash: None,
+                attribution: "lane.failed reconciled to lane.finished".to_string(),
+            }],
+        })
+    }
+
+    fn capabilities(families: &[&str], max_sensitivity: SensitivityClass) -> ConsumerCapabilities {
+        ConsumerCapabilities {
+            consumer: "clawhip".to_string(),
+            schema_versions: [REPORT_SCHEMA_V1.to_string()].into_iter().collect(),
+            field_families: families
+                .iter()
+                .map(|family| (*family).to_string())
+                .collect(),
+            max_sensitivity,
+        }
+    }
+
+    #[test]
+    fn report_schema_registry_is_self_describing() {
+        let registry = report_schema_v1_registry();
+        assert_eq!(registry.schema_version, REPORT_SCHEMA_V1);
+        assert!(registry
+            .fields
+            .iter()
+            .any(|field| field.id == "claims[].kind"));
+        assert!(registry
+            .fields
+            .iter()
+            .any(|field| field.id == "negative_evidence[]"));
+        assert!(registry
+            .fields
+            .iter()
+            .any(|field| field.id == "projection.provenance.redactions[]"));
+    }
+
+    #[test]
+    fn canonical_report_labels_claims_negative_evidence_and_deltas() {
+        let report = fixture_report();
+        assert_eq!(report.schema_version, REPORT_SCHEMA_V1);
+        assert!(report.identity.report_id.starts_with("report-"));
+        assert_eq!(report.identity.content_hash.len(), 16);
+        assert_eq!(report.claims[0].id, "claim-fact");
+        assert_eq!(report.claims[1].kind, ClaimKind::Hypothesis);
+        assert_eq!(report.claims[1].confidence, ReportConfidence::Medium);
+        assert_eq!(
+            report.negative_evidence[0].status,
+            NegativeFindingStatus::NotObservedInCheckedScope
+        );
+        assert_eq!(report.field_deltas[0].state, FieldDeltaState::Cleared);
+    }
+
+    #[test]
+    fn projections_are_deterministic_and_record_redaction_provenance() {
+        let report = fixture_report();
+        let capabilities = capabilities(
+            &["claims", "negative_evidence", "field_deltas"],
+            SensitivityClass::Public,
+        );
+
+        let first = project_report(&report, &capabilities, "delta_brief");
+        let second = project_report(&report, &capabilities, "delta_brief");
+
+        assert_eq!(first, second);
+        assert_eq!(first.provenance.source_report_id, report.identity.report_id);
+        assert_eq!(
+            first.provenance.source_content_hash,
+            report.identity.content_hash
+        );
+        assert!(first.provenance.downgraded);
+        assert_eq!(first.provenance.redactions.len(), 2);
+        assert!(first
+            .provenance
+            .redactions
+            .iter()
+            .any(|redaction| redaction.field_path == "claims[1].text"));
+        assert!(first
+            .provenance
+            .redactions
+            .iter()
+            .any(|redaction| redaction.field_path == "claims[2]"));
+    }
+
+    #[test]
+    fn capability_negotiation_omits_unsupported_field_families() {
+        let report = fixture_report();
+        let capabilities = capabilities(&["claims"], SensitivityClass::Internal);
+        let projection = project_report(&report, &capabilities, "legacy_clawhip");
+
+        assert!(projection.provenance.downgraded);
+        assert_eq!(
+            projection.provenance.omitted_field_families,
+            vec!["negative_evidence".to_string(), "field_deltas".to_string()]
+        );
+        assert!(projection.payload.get("claims").is_some());
+        assert!(projection.payload.get("negative_evidence").is_none());
+        assert!(projection.payload.get("field_deltas").is_none());
+    }
+}

rust/crates/runtime/src/sandbox.rs

@@ -298,8 +298,7 @@ fn unshare_user_namespace_works() -> bool {
             .stdout(std::process::Stdio::null())
             .stderr(std::process::Stdio::null())
             .status()
-            .map(|s| s.success())
+            .is_ok_and(|status| status.success())
-            .unwrap_or(false)
     })
 }
 

rust/crates/runtime/src/session.rs

@@ -30,6 +30,10 @@ pub enum ContentBlock {
     Text {
         text: String,
     },
+    Thinking {
+        thinking: String,
+        signature: Option<String>,
+    },
     ToolUse {
         id: String,
         name: String,
@@ -737,6 +741,22 @@ impl ContentBlock {
                 object.insert("type".to_string(), JsonValue::String("text".to_string()));
                 object.insert("text".to_string(), JsonValue::String(text.clone()));
             }
+            Self::Thinking {
+                thinking,
+                signature,
+            } => {
+                object.insert(
+                    "type".to_string(),
+                    JsonValue::String("thinking".to_string()),
+                );
+                object.insert("thinking".to_string(), JsonValue::String(thinking.clone()));
+                if let Some(signature) = signature {
+                    object.insert(
+                        "signature".to_string(),
+                        JsonValue::String(signature.clone()),
+                    );
+                }
+            }
             Self::ToolUse { id, name, input } => {
                 object.insert(
                     "type".to_string(),
@@ -783,6 +803,13 @@ impl ContentBlock {
             "text" => Ok(Self::Text {
                 text: required_string(object, "text")?,
             }),
+            "thinking" => Ok(Self::Thinking {
+                thinking: required_string(object, "thinking")?,
+                signature: object
+                    .get("signature")
+                    .and_then(JsonValue::as_str)
+                    .map(String::from),
+            }),
             "tool_use" => Ok(Self::ToolUse {
                 id: required_string(object, "id")?,
                 name: required_string(object, "name")?,
@@ -1208,6 +1235,36 @@ mod tests {
         assert_eq!(restored.session_id, session.session_id);
     }
 
+    #[test]
+    fn persists_assistant_thinking_block_round_trip_through_jsonl() {
+        // given
+        let mut session = Session::new();
+        session
+            .push_message(ConversationMessage::assistant(vec![
+                ContentBlock::Thinking {
+                    thinking: "trace the path through session persistence".to_string(),
+                    signature: Some("sig-123".to_string()),
+                },
+            ]))
+            .expect("thinking block should append");
+        let path = temp_session_path("thinking-jsonl");
+
+        // when
+        session.save_to_path(&path).expect("session should save");
+        let restored = Session::load_from_path(&path).expect("session should load");
+        fs::remove_file(&path).expect("temp file should be removable");
+
+        // then
+        assert_eq!(restored, session);
+        assert_eq!(
+            restored.messages[0].blocks[0],
+            ContentBlock::Thinking {
+                thinking: "trace the path through session persistence".to_string(),
+                signature: Some("sig-123".to_string()),
+            }
+        );
+    }
+
     #[test]
     fn loads_legacy_session_json_object() {
         let path = temp_session_path("legacy");

rust/crates/runtime/src/session_control.rs

@@ -58,8 +58,8 @@ impl SessionStore {
         let workspace_root = workspace_root.as_ref();
         // #151: canonicalize workspace_root for consistent fingerprinting
         // across equivalent path representations.
-        let canonical_workspace = fs::canonicalize(workspace_root)
+        let canonical_workspace =
-            .unwrap_or_else(|_| workspace_root.to_path_buf());
+            fs::canonicalize(workspace_root).unwrap_or_else(|_| workspace_root.to_path_buf());
         let sessions_root = data_dir
             .as_ref()
             .join("sessions")
@@ -158,10 +158,9 @@ impl SessionStore {
     }
 
     pub fn latest_session(&self) -> Result<ManagedSessionSummary, SessionControlError> {
-        self.list_sessions()?
+        self.list_sessions()?.into_iter().next().ok_or_else(|| {
-            .into_iter()
+            SessionControlError::Format(format_no_managed_sessions(&self.sessions_root))
-            .next()
+        })
-            .ok_or_else(|| SessionControlError::Format(format_no_managed_sessions(&self.sessions_root)))
     }
 
     pub fn load_session(

rust/crates/runtime/src/trust_resolver.rs

@@ -1,5 +1,7 @@
 use std::path::{Path, PathBuf};
 
+use serde::{Deserialize, Serialize};
+
 const TRUST_PROMPT_CUES: &[&str] = &[
     "do you trust the files in this folder",
     "trust the files in this folder",
@@ -8,24 +10,121 @@ const TRUST_PROMPT_CUES: &[&str] = &[
     "yes, proceed",
 ];
 
-#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+/// Resolution method for trust decisions.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
 pub enum TrustPolicy {
+    /// Automatically trust this path (allowlisted)
     AutoTrust,
+    /// Require manual approval
     RequireApproval,
+    /// Deny trust for this path
     Deny,
 }
 
-#[derive(Debug, Clone, PartialEq, Eq)]
+/// Events emitted during trust resolution lifecycle.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(tag = "type", rename_all = "snake_case")]
 pub enum TrustEvent {
-    TrustRequired { cwd: String },
+    /// Trust prompt was detected and is required
-    TrustResolved { cwd: String, policy: TrustPolicy },
+    TrustRequired {
-    TrustDenied { cwd: String, reason: String },
+        /// Current working directory where trust is needed
+        cwd: String,
+        /// Optional repo identifier
+        #[serde(skip_serializing_if = "Option::is_none")]
+        repo: Option<String>,
+        /// Optional worktree path
+        #[serde(skip_serializing_if = "Option::is_none")]
+        worktree: Option<String>,
+    },
+    /// Trust was resolved (granted)
+    TrustResolved {
+        /// Current working directory
+        cwd: String,
+        /// The policy that was applied
+        policy: TrustPolicy,
+        /// How the trust was resolved
+        resolution: TrustResolution,
+    },
+    /// Trust was denied
+    TrustDenied {
+        /// Current working directory
+        cwd: String,
+        /// Reason for denial
+        reason: String,
+    },
 }
 
-#[derive(Debug, Clone, Default)]
+/// How trust was resolved.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
+#[serde(rename_all = "snake_case")]
+pub enum TrustResolution {
+    /// Automatically granted due to allowlist
+    AutoAllowlisted,
+    /// Manually approved by user
+    ManualApproval,
+}
+
+/// Entry in the trust allowlist with pattern matching support.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct TrustAllowlistEntry {
+    /// Repository path or glob pattern to match
+    pub pattern: String,
+    /// Optional worktree subpath pattern
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub worktree_pattern: Option<String>,
+    /// Human-readable description of why this is allowlisted
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub description: Option<String>,
+}
+
+impl TrustAllowlistEntry {
+    #[must_use]
+    pub fn new(pattern: impl Into<String>) -> Self {
+        Self {
+            pattern: pattern.into(),
+            worktree_pattern: None,
+            description: None,
+        }
+    }
+
+    #[must_use]
+    pub fn with_worktree_pattern(mut self, pattern: impl Into<String>) -> Self {
+        self.worktree_pattern = Some(pattern.into());
+        self
+    }
+
+    #[must_use]
+    pub fn with_description(mut self, desc: impl Into<String>) -> Self {
+        self.description = Some(desc.into());
+        self
+    }
+}
+
+/// Configuration for trust resolution with allowlist/denylist support.
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
 pub struct TrustConfig {
-    allowlisted: Vec<PathBuf>,
+    /// Allowlisted paths with pattern matching
-    denied: Vec<PathBuf>,
+    pub allowlisted: Vec<TrustAllowlistEntry>,
+    /// Denied paths (exact or prefix matches)
+    pub denied: Vec<PathBuf>,
+    /// Whether to emit events for trust decisions
+    #[serde(default = "default_emit_events")]
+    pub emit_events: bool,
+}
+
+fn default_emit_events() -> bool {
+    true
+}
+
+impl Default for TrustConfig {
+    fn default() -> Self {
+        Self {
+            allowlisted: Vec::new(),
+            denied: Vec::new(),
+            emit_events: true,
+        }
+    }
 }
 
 impl TrustConfig {
@@ -35,8 +134,14 @@ impl TrustConfig {
     }
 
     #[must_use]
-    pub fn with_allowlisted(mut self, path: impl Into<PathBuf>) -> Self {
+    pub fn with_allowlisted(mut self, path: impl Into<String>) -> Self {
-        self.allowlisted.push(path.into());
+        self.allowlisted.push(TrustAllowlistEntry::new(path));
+        self
+    }
+
+    #[must_use]
+    pub fn with_allowlisted_entry(mut self, entry: TrustAllowlistEntry) -> Self {
+        self.allowlisted.push(entry);
         self
     }
 
@@ -45,6 +150,147 @@ impl TrustConfig {
         self.denied.push(path.into());
         self
     }
+
+    /// Check if a path matches an allowlisted entry using glob patterns.
+    #[must_use]
+    pub fn is_allowlisted(
+        &self,
+        cwd: &str,
+        worktree: Option<&str>,
+    ) -> Option<&TrustAllowlistEntry> {
+        self.allowlisted.iter().find(|entry| {
+            let path_matches = Self::pattern_matches(&entry.pattern, cwd);
+            if !path_matches {
+                return false;
+            }
+
+            match (&entry.worktree_pattern, worktree) {
+                (Some(wt_pattern), Some(wt)) => Self::pattern_matches(wt_pattern, wt),
+                (Some(_), None) => false,
+                (None, _) => true,
+            }
+        })
+    }
+
+    /// Match a pattern against a path string.
+    /// Supports exact matching and glob patterns (* and ?).
+    fn pattern_matches(pattern: &str, path: &str) -> bool {
+        let pattern = pattern.trim();
+        let path = path.trim();
+
+        // Exact match
+        if pattern == path {
+            return true;
+        }
+
+        // Normalize paths for comparison
+        let pattern_normalized = pattern.replace("//", "/");
+        let path_normalized = path.replace("//", "/");
+
+        // Check if pattern is a path prefix (e.g., "/tmp/worktrees" matches "/tmp/worktrees/repo-a")
+        // This handles the common case of directory containment
+        if !pattern_normalized.contains('*') && !pattern_normalized.contains('?') {
+            // Prefix match: pattern is a directory that contains path
+            if path_normalized.starts_with(&pattern_normalized) {
+                let rest = &path_normalized[pattern_normalized.len()..];
+                // Must be exact match or continue with /
+                return rest.is_empty() || rest.starts_with('/');
+            }
+        }
+
+        // Check if pattern ends with wildcard (prefix match)
+        if pattern_normalized.ends_with("/*") {
+            let prefix = pattern_normalized.trim_end_matches("/*");
+            if let Some(rest) = path_normalized.strip_prefix(prefix) {
+                // Must either be exact match or continue with /
+                return rest.is_empty() || rest.starts_with('/');
+            }
+        } else if pattern_normalized.ends_with('*') && !pattern_normalized.contains("/*/") {
+            // Simple trailing * (not a path component wildcard)
+            let prefix = pattern_normalized.trim_end_matches('*');
+            if let Some(rest) = path_normalized.strip_prefix(prefix) {
+                return rest.is_empty() || !rest.starts_with('/');
+            }
+        }
+
+        // Check if pattern is a path component match (bounded by /)
+        if path_normalized
+            .split('/')
+            .any(|component| component == pattern_normalized)
+        {
+            return true;
+        }
+
+        // Check if pattern appears as a substring within a path component
+        // (e.g., "repo" matches "/tmp/worktrees/repo-a")
+        if path_normalized
+            .split('/')
+            .any(|component| component.contains(&pattern_normalized))
+        {
+            return true;
+        }
+
+        // Glob matching for patterns with ? or * in the middle
+        if pattern.contains('?') || pattern.contains("/*/") || pattern.starts_with("*/") {
+            return Self::glob_matches(&pattern_normalized, &path_normalized);
+        }
+
+        false
+    }
+
+    /// Simple glob pattern matching (? matches single char, * matches any sequence).
+    /// Handles patterns like /tmp/*/repo-* where * matches path components.
+    fn glob_matches(pattern: &str, path: &str) -> bool {
+        // Use recursive backtracking for proper glob matching
+        Self::glob_match_recursive(pattern, path, 0, 0)
+    }
+
+    fn glob_match_recursive(pattern: &str, path: &str, p_idx: usize, s_idx: usize) -> bool {
+        let p_chars: Vec<char> = pattern.chars().collect();
+        let s_chars: Vec<char> = path.chars().collect();
+
+        let mut p = p_idx;
+        let mut s = s_idx;
+
+        while p < p_chars.len() {
+            match p_chars[p] {
+                '*' => {
+                    // Try all possible matches for *
+                    p += 1;
+                    if p >= p_chars.len() {
+                        // * at end matches everything remaining
+                        return true;
+                    }
+                    // Try matching 0 or more characters
+                    for skip in 0..=(s_chars.len() - s) {
+                        if Self::glob_match_recursive(pattern, path, p, s + skip) {
+                            return true;
+                        }
+                    }
+                    return false;
+                }
+                '?' => {
+                    // ? matches exactly one character
+                    if s >= s_chars.len() {
+                        return false;
+                    }
+                    p += 1;
+                    s += 1;
+                }
+                c => {
+                    // Exact character match
+                    if s >= s_chars.len() || s_chars[s] != c {
+                        return false;
+                    }
+                    p += 1;
+                    s += 1;
+                }
+            }
+        }
+
+        // Pattern exhausted - path must also be exhausted
+        s >= s_chars.len()
+    }
 }
 
 #[derive(Debug, Clone, PartialEq, Eq)]
@@ -86,15 +332,19 @@ impl TrustResolver {
     }
 
     #[must_use]
-    pub fn resolve(&self, cwd: &str, screen_text: &str) -> TrustDecision {
+    pub fn resolve(&self, cwd: &str, worktree: Option<&str>, screen_text: &str) -> TrustDecision {
         if !detect_trust_prompt(screen_text) {
             return TrustDecision::NotRequired;
         }
 
+        let repo = extract_repo_name(cwd);
         let mut events = vec![TrustEvent::TrustRequired {
             cwd: cwd.to_owned(),
+            repo: repo.clone(),
+            worktree: worktree.map(String::from),
         }];
 
+        // Check denylist first
         if let Some(matched_root) = self
             .config
             .denied
@@ -112,15 +362,12 @@ impl TrustResolver {
             };
         }
 
-        if self
+        // Check allowlist with pattern matching
-            .config
+        if self.config.is_allowlisted(cwd, worktree).is_some() {
-            .allowlisted
-            .iter()
-            .any(|root| path_matches(cwd, root))
-        {
             events.push(TrustEvent::TrustResolved {
                 cwd: cwd.to_owned(),
                 policy: TrustPolicy::AutoTrust,
+                resolution: TrustResolution::AutoAllowlisted,
             });
             return TrustDecision::Required {
                 policy: TrustPolicy::AutoTrust,
@@ -128,6 +375,19 @@ impl TrustResolver {
             };
         }
 
+        // Check for manual trust resolution via screen text analysis
+        if detect_manual_approval(screen_text) {
+            events.push(TrustEvent::TrustResolved {
+                cwd: cwd.to_owned(),
+                policy: TrustPolicy::RequireApproval,
+                resolution: TrustResolution::ManualApproval,
+            });
+            return TrustDecision::Required {
+                policy: TrustPolicy::RequireApproval,
+                events,
+            };
+        }
+
         TrustDecision::Required {
             policy: TrustPolicy::RequireApproval,
             events,
@@ -135,17 +395,20 @@ impl TrustResolver {
     }
 
     #[must_use]
-    pub fn trusts(&self, cwd: &str) -> bool {
+    pub fn trusts(&self, cwd: &str, worktree: Option<&str>) -> bool {
-        !self
+        // Check denylist first
+        let denied = self
             .config
             .denied
             .iter()
-            .any(|root| path_matches(cwd, root))
+            .any(|root| path_matches(cwd, root));
-            && self
+
-                .config
+        if denied {
-                .allowlisted
+            return false;
-                .iter()
+        }
-                .any(|root| path_matches(cwd, root))
+
+        // Check allowlist using pattern matching
+        self.config.is_allowlisted(cwd, worktree).is_some()
     }
 }
 
@@ -172,11 +435,240 @@ fn normalize_path(path: &Path) -> PathBuf {
     std::fs::canonicalize(path).unwrap_or_else(|_| path.to_path_buf())
 }
 
+/// Extract repository name from a path for event context.
+fn extract_repo_name(cwd: &str) -> Option<String> {
+    let path = Path::new(cwd);
+    // Try to find a .git directory to identify repo root
+    let mut current = Some(path);
+    while let Some(p) = current {
+        if p.join(".git").is_dir() {
+            return p.file_name().map(|n| n.to_string_lossy().to_string());
+        }
+        current = p.parent();
+    }
+    // Fallback: use the last component of the path
+    path.file_name().map(|n| n.to_string_lossy().to_string())
+}
+
+/// Detect if the screen text indicates manual approval was granted.
+fn detect_manual_approval(screen_text: &str) -> bool {
+    let lowered = screen_text.to_ascii_lowercase();
+    // Look for indicators that user manually approved
+    MANUAL_APPROVAL_CUES.iter().any(|cue| lowered.contains(cue))
+}
+
+const MANUAL_APPROVAL_CUES: &[&str] = &[
+    "yes, i trust",
+    "i trust this",
+    "trusted manually",
+    "approval granted",
+];
+
+#[cfg(test)]
+mod path_matching_tests {
+    use super::*;
+
+    #[test]
+    fn glob_pattern_star_matches_any_sequence() {
+        assert!(TrustConfig::pattern_matches("/tmp/*", "/tmp/foo"));
+        assert!(TrustConfig::pattern_matches("/tmp/*", "/tmp/bar/baz"));
+        assert!(!TrustConfig::pattern_matches("/tmp/*", "/other/tmp/foo"));
+    }
+
+    #[test]
+    fn glob_pattern_question_matches_single_char() {
+        assert!(TrustConfig::pattern_matches("/tmp/test?", "/tmp/test1"));
+        assert!(TrustConfig::pattern_matches("/tmp/test?", "/tmp/testA"));
+        assert!(!TrustConfig::pattern_matches("/tmp/test?", "/tmp/test12"));
+        assert!(!TrustConfig::pattern_matches("/tmp/test?", "/tmp/test"));
+    }
+
+    #[test]
+    fn pattern_matches_exact() {
+        assert!(TrustConfig::pattern_matches(
+            "/tmp/worktrees",
+            "/tmp/worktrees"
+        ));
+        assert!(!TrustConfig::pattern_matches(
+            "/tmp/worktrees",
+            "/tmp/worktrees-other"
+        ));
+    }
+
+    #[test]
+    fn pattern_matches_prefix_with_wildcard() {
+        assert!(TrustConfig::pattern_matches(
+            "/tmp/worktrees/*",
+            "/tmp/worktrees/repo-a"
+        ));
+        assert!(TrustConfig::pattern_matches(
+            "/tmp/worktrees/*",
+            "/tmp/worktrees/repo-a/subdir"
+        ));
+        assert!(!TrustConfig::pattern_matches(
+            "/tmp/worktrees/*",
+            "/tmp/other/repo"
+        ));
+    }
+
+    #[test]
+    fn pattern_matches_contains() {
+        // Pattern contained within path
+        assert!(TrustConfig::pattern_matches(
+            "worktrees",
+            "/tmp/worktrees/repo-a"
+        ));
+        assert!(TrustConfig::pattern_matches(
+            "repo",
+            "/tmp/worktrees/repo-a"
+        ));
+    }
+
+    #[test]
+    fn allowlist_entry_with_worktree_pattern() {
+        let config = TrustConfig::new().with_allowlisted_entry(
+            TrustAllowlistEntry::new("/tmp/worktrees/*")
+                .with_worktree_pattern("*/.git")
+                .with_description("Git worktrees"),
+        );
+
+        // Should match when both patterns match
+        assert!(config
+            .is_allowlisted("/tmp/worktrees/repo-a", Some("/tmp/worktrees/repo-a/.git"))
+            .is_some());
+
+        // Should not match when worktree pattern doesn't match
+        assert!(config
+            .is_allowlisted("/tmp/worktrees/repo-a", Some("/other/path"))
+            .is_none());
+
+        // Should not match when a worktree pattern is required but no worktree is supplied
+        assert!(config
+            .is_allowlisted("/tmp/worktrees/repo-a", None)
+            .is_none());
+
+        // Should match when no worktree pattern required and path matches
+        let config_no_worktree = TrustConfig::new().with_allowlisted("/tmp/worktrees/*");
+        assert!(config_no_worktree
+            .is_allowlisted("/tmp/worktrees/repo-a", None)
+            .is_some());
+    }
+
+    #[test]
+    fn allowlist_entry_returns_matched_entry() {
+        let entry = TrustAllowlistEntry::new("/tmp/worktrees/*").with_description("Test worktrees");
+        let config = TrustConfig::new().with_allowlisted_entry(entry.clone());
+
+        let matched = config.is_allowlisted("/tmp/worktrees/repo-a", None);
+        assert!(matched.is_some());
+        assert_eq!(
+            matched.unwrap().description,
+            Some("Test worktrees".to_string())
+        );
+    }
+
+    #[test]
+    fn complex_glob_patterns() {
+        // Multiple wildcards
+        assert!(TrustConfig::pattern_matches(
+            "/tmp/*/repo-*",
+            "/tmp/worktrees/repo-123"
+        ));
+        assert!(TrustConfig::pattern_matches(
+            "/tmp/*/repo-*",
+            "/tmp/other/repo-abc"
+        ));
+        assert!(!TrustConfig::pattern_matches(
+            "/tmp/*/repo-*",
+            "/tmp/worktrees/other"
+        ));
+
+        // Mixed ? and *
+        assert!(TrustConfig::pattern_matches(
+            "/tmp/test?/*.txt",
+            "/tmp/test1/file.txt"
+        ));
+        assert!(TrustConfig::pattern_matches(
+            "/tmp/test?/*.txt",
+            "/tmp/testA/subdir/file.txt"
+        ));
+    }
+
+    #[test]
+    fn serde_serialization_roundtrip() {
+        let config = TrustConfig::new()
+            .with_allowlisted_entry(
+                TrustAllowlistEntry::new("/tmp/worktrees/*")
+                    .with_worktree_pattern("*/.git")
+                    .with_description("Git worktrees"),
+            )
+            .with_denied("/tmp/malicious");
+
+        let json = serde_json::to_string(&config).expect("serialization failed");
+        let deserialized: TrustConfig =
+            serde_json::from_str(&json).expect("deserialization failed");
+
+        assert_eq!(config.allowlisted.len(), deserialized.allowlisted.len());
+        assert_eq!(config.denied.len(), deserialized.denied.len());
+        assert_eq!(config.emit_events, deserialized.emit_events);
+    }
+
+    #[test]
+    fn trust_event_serialization() {
+        let event = TrustEvent::TrustRequired {
+            cwd: "/tmp/test".to_string(),
+            repo: Some("test-repo".to_string()),
+            worktree: Some("/tmp/test/.git".to_string()),
+        };
+
+        let json = serde_json::to_string(&event).expect("serialization failed");
+        assert!(json.contains("trust_required"));
+        assert!(json.contains("/tmp/test"));
+        assert!(json.contains("test-repo"));
+
+        let deserialized: TrustEvent = serde_json::from_str(&json).expect("deserialization failed");
+        match deserialized {
+            TrustEvent::TrustRequired {
+                cwd,
+                repo,
+                worktree,
+            } => {
+                assert_eq!(cwd, "/tmp/test");
+                assert_eq!(repo, Some("test-repo".to_string()));
+                assert_eq!(worktree, Some("/tmp/test/.git".to_string()));
+            }
+            _ => panic!("wrong event type"),
+        }
+    }
+
+    #[test]
+    fn trust_event_resolved_serialization() {
+        let event = TrustEvent::TrustResolved {
+            cwd: "/tmp/test".to_string(),
+            policy: TrustPolicy::AutoTrust,
+            resolution: TrustResolution::AutoAllowlisted,
+        };
+
+        let json = serde_json::to_string(&event).expect("serialization failed");
+        assert!(json.contains("trust_resolved"));
+        assert!(json.contains("auto_allowlisted"));
+
+        let deserialized: TrustEvent = serde_json::from_str(&json).expect("deserialization failed");
+        match deserialized {
+            TrustEvent::TrustResolved { resolution, .. } => {
+                assert_eq!(resolution, TrustResolution::AutoAllowlisted);
+            }
+            _ => panic!("wrong event type"),
+        }
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::{
-        detect_trust_prompt, path_matches_trusted_root, TrustConfig, TrustDecision, TrustEvent,
+        detect_manual_approval, detect_trust_prompt, path_matches_trusted_root,
-        TrustPolicy, TrustResolver,
+        TrustAllowlistEntry, TrustConfig, TrustDecision, TrustEvent, TrustPolicy, TrustResolution,
+        TrustResolver,
     };
 
     #[test]
@@ -197,7 +689,7 @@ mod tests {
         let resolver = TrustResolver::new(TrustConfig::new().with_allowlisted("/tmp/worktrees"));
 
         // when
-        let decision = resolver.resolve("/tmp/worktrees/repo-a", "Ready for your input\n>");
+        let decision = resolver.resolve("/tmp/worktrees/repo-a", None, "Ready for your input\n>");
 
         // then
         assert_eq!(decision, TrustDecision::NotRequired);
@@ -213,23 +705,23 @@ mod tests {
         // when
         let decision = resolver.resolve(
             "/tmp/worktrees/repo-a",
+            None,
             "Do you trust the files in this folder?\n1. Yes, proceed\n2. No",
         );
 
         // then
         assert_eq!(decision.policy(), Some(TrustPolicy::AutoTrust));
-        assert_eq!(
+        let events = decision.events();
-            decision.events(),
+        assert_eq!(events.len(), 2);
-            &[
+        assert!(matches!(events[0], TrustEvent::TrustRequired { .. }));
-                TrustEvent::TrustRequired {
+        assert!(matches!(
-                    cwd: "/tmp/worktrees/repo-a".to_string(),
+            events[1],
-                },
+            TrustEvent::TrustResolved {
-                TrustEvent::TrustResolved {
+                policy: TrustPolicy::AutoTrust,
-                    cwd: "/tmp/worktrees/repo-a".to_string(),
+                resolution: TrustResolution::AutoAllowlisted,
-                    policy: TrustPolicy::AutoTrust,
+                ..
-                },
+            }
-            ]
+        ));
-        );
     }
 
     #[test]
@@ -240,6 +732,7 @@ mod tests {
         // when
         let decision = resolver.resolve(
             "/tmp/other/repo-b",
+            None,
             "Do you trust the files in this folder?\n1. Yes, proceed\n2. No",
         );
 
@@ -249,6 +742,8 @@ mod tests {
             decision.events(),
             &[TrustEvent::TrustRequired {
                 cwd: "/tmp/other/repo-b".to_string(),
+                repo: Some("repo-b".to_string()),
+                worktree: None,
             }]
         );
     }
@@ -265,6 +760,7 @@ mod tests {
         // when
         let decision = resolver.resolve(
             "/tmp/worktrees/repo-c",
+            None,
             "Do you trust the files in this folder?\n1. Yes, proceed\n2. No",
         );
 
@@ -275,6 +771,8 @@ mod tests {
             &[
                 TrustEvent::TrustRequired {
                     cwd: "/tmp/worktrees/repo-c".to_string(),
+                    repo: Some("repo-c".to_string()),
+                    worktree: None,
                 },
                 TrustEvent::TrustDenied {
                     cwd: "/tmp/worktrees/repo-c".to_string(),
@@ -284,6 +782,66 @@ mod tests {
         );
     }
 
+    #[test]
+    fn auto_trusts_with_glob_pattern_allowlist() {
+        // given
+        let resolver = TrustResolver::new(TrustConfig::new().with_allowlisted("/tmp/worktrees/*"));
+
+        // when - any repo under /tmp/worktrees should auto-trust
+        let decision = resolver.resolve(
+            "/tmp/worktrees/repo-a",
+            None,
+            "Do you trust the files in this folder?\n1. Yes, proceed\n2. No",
+        );
+
+        // then
+        assert_eq!(decision.policy(), Some(TrustPolicy::AutoTrust));
+    }
+
+    #[test]
+    fn resolve_with_worktree_pattern_matching() {
+        // given
+        let config = TrustConfig::new().with_allowlisted_entry(
+            TrustAllowlistEntry::new("/tmp/worktrees/*").with_worktree_pattern("*/.git"),
+        );
+        let resolver = TrustResolver::new(config);
+
+        // when - with worktree that matches the pattern
+        let decision = resolver.resolve(
+            "/tmp/worktrees/repo-a",
+            Some("/tmp/worktrees/repo-a/.git"),
+            "Do you trust the files in this folder?\n1. Yes, proceed\n2. No",
+        );
+
+        // then - should auto-trust because both patterns match
+        assert_eq!(decision.policy(), Some(TrustPolicy::AutoTrust));
+    }
+
+    #[test]
+    fn manual_approval_detected_from_screen_text() {
+        // given
+        let resolver = TrustResolver::new(TrustConfig::new());
+
+        // when - screen text indicates manual approval
+        let decision = resolver.resolve(
+            "/tmp/some/repo",
+            None,
+            "Do you trust the files in this folder?\nUser selected: Yes, I trust this folder",
+        );
+
+        // then - should detect manual approval
+        assert_eq!(decision.policy(), Some(TrustPolicy::RequireApproval));
+        let events = decision.events();
+        assert!(events.len() >= 2);
+        assert!(matches!(
+            events[events.len() - 1],
+            TrustEvent::TrustResolved {
+                resolution: TrustResolution::ManualApproval,
+                ..
+            }
+        ));
+    }
+
     #[test]
     fn sibling_prefix_does_not_match_trusted_root() {
         // given
@@ -296,4 +854,70 @@ mod tests {
         // then
         assert!(!matched);
     }
+
+    #[test]
+    fn detects_manual_approval_cues() {
+        assert!(detect_manual_approval(
+            "User selected: Yes, I trust this folder"
+        ));
+        assert!(detect_manual_approval(
+            "I trust this repository and its contents"
+        ));
+        assert!(detect_manual_approval("Approval granted by user"));
+        assert!(!detect_manual_approval(
+            "Do you trust the files in this folder?"
+        ));
+        assert!(!detect_manual_approval("Some unrelated text"));
+    }
+
+    #[test]
+    fn trust_config_default_emit_events() {
+        let config = TrustConfig::default();
+        assert!(config.emit_events);
+    }
+
+    #[test]
+    fn trust_resolver_trusts_method() {
+        let resolver = TrustResolver::new(
+            TrustConfig::new()
+                .with_allowlisted("/tmp/worktrees/*")
+                .with_denied("/tmp/worktrees/bad-repo"),
+        );
+
+        // Should trust allowlisted paths
+        assert!(resolver.trusts("/tmp/worktrees/good-repo", None));
+
+        // Should not trust denied paths
+        assert!(!resolver.trusts("/tmp/worktrees/bad-repo", None));
+
+        // Should not trust unknown paths
+        assert!(!resolver.trusts("/tmp/other/repo", None));
+    }
+
+    #[test]
+    fn trust_policy_serde_roundtrip() {
+        for policy in [
+            TrustPolicy::AutoTrust,
+            TrustPolicy::RequireApproval,
+            TrustPolicy::Deny,
+        ] {
+            let json = serde_json::to_string(&policy).expect("serialization failed");
+            let deserialized: TrustPolicy =
+                serde_json::from_str(&json).expect("deserialization failed");
+            assert_eq!(policy, deserialized);
+        }
+    }
+
+    #[test]
+    fn trust_resolution_serde_roundtrip() {
+        for resolution in [
+            TrustResolution::AutoAllowlisted,
+            TrustResolution::ManualApproval,
+        ] {
+            let json = serde_json::to_string(&resolution).expect("serialization failed");
+            let deserialized: TrustResolution =
+                serde_json::from_str(&json).expect("deserialization failed");
+            assert_eq!(resolution, deserialized);
+        }
+    }
 }

rust/crates/runtime/src/worker_boot.rs

@@ -30,6 +30,7 @@ fn now_secs() -> u64 {
 pub enum WorkerStatus {
     Spawning,
     TrustRequired,
+    ToolPermissionRequired,
     ReadyForPrompt,
     Running,
     Finished,
@@ -41,6 +42,7 @@ impl std::fmt::Display for WorkerStatus {
         match self {
             Self::Spawning => write!(f, "spawning"),
             Self::TrustRequired => write!(f, "trust_required"),
+            Self::ToolPermissionRequired => write!(f, "tool_permission_required"),
             Self::ReadyForPrompt => write!(f, "ready_for_prompt"),
             Self::Running => write!(f, "running"),
             Self::Finished => write!(f, "finished"),
@@ -53,6 +55,7 @@ impl std::fmt::Display for WorkerStatus {
 #[serde(rename_all = "snake_case")]
 pub enum WorkerFailureKind {
     TrustGate,
+    ToolPermissionGate,
     PromptDelivery,
     Protocol,
     Provider,
@@ -71,6 +74,7 @@ pub struct WorkerFailure {
 pub enum WorkerEventKind {
     Spawning,
     TrustRequired,
+    ToolPermissionRequired,
     TrustResolved,
     ReadyForPrompt,
     PromptMisdelivery,
@@ -104,6 +108,8 @@ pub enum WorkerPromptTarget {
 pub enum StartupFailureClassification {
     /// Trust prompt is required but not detected/resolved
     TrustRequired,
+    /// Tool permission prompt is required before startup can continue
+    ToolPermissionRequired,
     /// Prompt was delivered to wrong target (shell misdelivery)
     PromptMisdelivery,
     /// Prompt was sent but acceptance timed out
@@ -116,13 +122,37 @@ pub enum StartupFailureClassification {
     Unknown,
 }
 
+#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
+pub struct StartupHealthSummary {
+    /// Whether this subsystem appeared healthy at timeout.
+    pub healthy: bool,
+    /// Stable placeholder/source string until deeper transport and MCP probes are wired in.
+    pub summary: String,
+}
+
+impl StartupHealthSummary {
+    fn observed(name: &str, healthy: bool) -> Self {
+        let status = if healthy { "healthy" } else { "unhealthy" };
+        Self {
+            healthy,
+            summary: format!("{name}_{status}_placeholder"),
+        }
+    }
+}
+
 /// Evidence bundle collected when worker startup times out without clear evidence.
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
 pub struct StartupEvidenceBundle {
     /// Last known worker lifecycle state before timeout
     pub last_lifecycle_state: WorkerStatus,
+    /// Timestamp of the last lifecycle state transition, unix epoch seconds
+    pub last_lifecycle_at: u64,
     /// The pane/command that was being executed
     pub pane_command: String,
+    /// Timestamp when the pane/command snapshot was observed, unix epoch seconds
+    pub pane_observed_at: u64,
+    /// Timestamp when the worker command was started, unix epoch seconds
+    pub command_started_at: u64,
     /// Timestamp when prompt was sent (if any), unix epoch seconds
     #[serde(skip_serializing_if = "Option::is_none")]
     pub prompt_sent_at: Option<u64>,
@@ -130,10 +160,22 @@ pub struct StartupEvidenceBundle {
     pub prompt_acceptance_state: bool,
     /// Result of trust prompt detection at timeout
     pub trust_prompt_detected: bool,
+    /// Result of tool permission prompt detection at timeout
+    pub tool_permission_prompt_detected: bool,
+    /// Age in seconds of the latest tool permission prompt, when observed
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub tool_permission_prompt_age_seconds: Option<u64>,
+    /// Whether the prompt surface exposed only a session allow path or also an always-allow path
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub tool_permission_allow_scope: Option<ToolPermissionAllowScope>,
     /// Transport health summary (true = healthy/responsive)
     pub transport_healthy: bool,
+    /// Typed transport health placeholder for future concrete probes
+    pub transport_health: StartupHealthSummary,
     /// MCP health summary (true = all servers healthy)
     pub mcp_healthy: bool,
+    /// Typed MCP health placeholder for future concrete probes
+    pub mcp_health: StartupHealthSummary,
     /// Seconds since worker creation
     pub elapsed_seconds: u64,
 }
@@ -146,6 +188,15 @@ pub enum WorkerEventPayload {
         #[serde(skip_serializing_if = "Option::is_none")]
         resolution: Option<WorkerTrustResolution>,
     },
+    ToolPermissionPrompt {
+        #[serde(skip_serializing_if = "Option::is_none")]
+        server_name: Option<String>,
+        #[serde(skip_serializing_if = "Option::is_none")]
+        tool_name: Option<String>,
+        prompt_age_seconds: u64,
+        allow_scope: ToolPermissionAllowScope,
+        prompt_preview: String,
+    },
     PromptDelivery {
         prompt_preview: String,
         observed_target: WorkerPromptTarget,
@@ -163,6 +214,14 @@ pub enum WorkerEventPayload {
     },
 }
 
+#[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq)]
+#[serde(rename_all = "snake_case")]
+pub enum ToolPermissionAllowScope {
+    SessionOnly,
+    SessionOrAlways,
+    Unknown,
+}
+
 #[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
 pub struct WorkerTaskReceipt {
     pub repo: String,
@@ -194,6 +253,7 @@ pub struct Worker {
     pub auto_recover_prompt_misdelivery: bool,
     pub prompt_delivery_attempts: u32,
     pub prompt_in_flight: bool,
+    pub prompt_sent_at: Option<u64>,
     pub last_prompt: Option<String>,
     pub expected_receipt: Option<WorkerTaskReceipt>,
     pub replay_prompt: Option<String>,
@@ -243,6 +303,7 @@ impl WorkerRegistry {
             auto_recover_prompt_misdelivery,
             prompt_delivery_attempts: 0,
             prompt_in_flight: false,
+            prompt_sent_at: None,
             last_prompt: None,
             expected_receipt: None,
             replay_prompt: None,
@@ -276,6 +337,29 @@ impl WorkerRegistry {
             .ok_or_else(|| format!("worker not found: {worker_id}"))?;
         let lowered = screen_text.to_ascii_lowercase();
 
+        if let Some(tool_prompt) = detect_tool_permission_prompt(screen_text, &lowered) {
+            worker.status = WorkerStatus::ToolPermissionRequired;
+            worker.last_error = Some(WorkerFailure {
+                kind: WorkerFailureKind::ToolPermissionGate,
+                message: tool_prompt.message(),
+                created_at: now_secs(),
+            });
+            push_event(
+                worker,
+                WorkerEventKind::ToolPermissionRequired,
+                WorkerStatus::ToolPermissionRequired,
+                Some("tool permission prompt detected".to_string()),
+                Some(WorkerEventPayload::ToolPermissionPrompt {
+                    server_name: tool_prompt.server_name,
+                    tool_name: tool_prompt.tool_name,
+                    prompt_age_seconds: 0,
+                    allow_scope: tool_prompt.allow_scope,
+                    prompt_preview: tool_prompt.prompt_preview,
+                }),
+            );
+            return Ok(worker.clone());
+        }
+
         if !worker.trust_gate_cleared && detect_trust_prompt(&lowered) {
             worker.status = WorkerStatus::TrustRequired;
             worker.last_error = Some(WorkerFailure {
@@ -474,6 +558,7 @@ impl WorkerRegistry {
 
         worker.prompt_delivery_attempts += 1;
         worker.prompt_in_flight = true;
+        worker.prompt_sent_at = Some(now_secs());
         worker.last_prompt = Some(next_prompt.clone());
         worker.expected_receipt = task_receipt;
         worker.replay_prompt = None;
@@ -503,7 +588,9 @@ impl WorkerRegistry {
             ready: worker.status == WorkerStatus::ReadyForPrompt,
             blocked: matches!(
                 worker.status,
-                WorkerStatus::TrustRequired | WorkerStatus::Failed
+                WorkerStatus::TrustRequired
+                    | WorkerStatus::ToolPermissionRequired
+                    | WorkerStatus::Failed
             ),
             replay_prompt_ready: worker.replay_prompt.is_some(),
             last_error: worker.last_error.clone(),
@@ -523,6 +610,7 @@ impl WorkerRegistry {
         worker.last_error = None;
         worker.prompt_delivery_attempts = 0;
         worker.prompt_in_flight = false;
+        worker.prompt_sent_at = None;
         push_event(
             worker,
             WorkerEventKind::Restarted,
@@ -624,24 +712,44 @@ impl WorkerRegistry {
 
         let now = now_secs();
         let elapsed = now.saturating_sub(worker.created_at);
+        let latest_tool_permission_event = worker
+            .events
+            .iter()
+            .rev()
+            .find(|event| event.kind == WorkerEventKind::ToolPermissionRequired);
+        let tool_permission_allow_scope =
+            latest_tool_permission_event.and_then(|event| match &event.payload {
+                Some(WorkerEventPayload::ToolPermissionPrompt { allow_scope, .. }) => {
+                    Some(*allow_scope)
+                }
+                _ => None,
+            });
 
         // Build evidence bundle
         let evidence = StartupEvidenceBundle {
             last_lifecycle_state: worker.status,
+            last_lifecycle_at: worker.updated_at,
             pane_command: pane_command.to_string(),
-            prompt_sent_at: if worker.prompt_delivery_attempts > 0 {
+            pane_observed_at: now,
-                Some(worker.updated_at)
+            command_started_at: worker.created_at,
-            } else {
+            prompt_sent_at: worker.prompt_sent_at,
-                None
-            },
             prompt_acceptance_state: worker.status == WorkerStatus::Running
                 && !worker.prompt_in_flight,
             trust_prompt_detected: worker
                 .events
                 .iter()
                 .any(|e| e.kind == WorkerEventKind::TrustRequired),
+            tool_permission_prompt_detected: worker
+                .events
+                .iter()
+                .any(|e| e.kind == WorkerEventKind::ToolPermissionRequired),
+            tool_permission_prompt_age_seconds: latest_tool_permission_event
+                .map(|event| now.saturating_sub(event.timestamp)),
+            tool_permission_allow_scope,
             transport_healthy,
+            transport_health: StartupHealthSummary::observed("transport", transport_healthy),
             mcp_healthy,
+            mcp_health: StartupHealthSummary::observed("mcp", mcp_healthy),
             elapsed_seconds: elapsed,
         };
 
@@ -694,6 +802,13 @@ fn classify_startup_failure(evidence: &StartupEvidenceBundle) -> StartupFailureC
         return StartupFailureClassification::TrustRequired;
     }
 
+    // Check for tool permission prompts that were not resolved
+    if evidence.tool_permission_prompt_detected
+        && evidence.last_lifecycle_state == WorkerStatus::ToolPermissionRequired
+    {
+        return StartupFailureClassification::ToolPermissionRequired;
+    }
+
     // Check for prompt acceptance timeout
     if evidence.prompt_sent_at.is_some()
         && !evidence.prompt_acceptance_state
@@ -815,6 +930,140 @@ fn normalize_path(path: &str) -> PathBuf {
     std::fs::canonicalize(path).unwrap_or_else(|_| Path::new(path).to_path_buf())
 }
 
+#[derive(Debug, Clone, PartialEq, Eq)]
+struct ToolPermissionPromptObservation {
+    server_name: Option<String>,
+    tool_name: Option<String>,
+    allow_scope: ToolPermissionAllowScope,
+    prompt_preview: String,
+}
+
+impl ToolPermissionPromptObservation {
+    fn message(&self) -> String {
+        match (&self.server_name, &self.tool_name) {
+            (Some(server), Some(tool)) => {
+                format!("worker boot blocked on tool permission prompt for {server}.{tool}")
+            }
+            (Some(server), None) => {
+                format!("worker boot blocked on tool permission prompt for {server}")
+            }
+            (None, Some(tool)) => {
+                format!("worker boot blocked on tool permission prompt for {tool}")
+            }
+            (None, None) => "worker boot blocked on tool permission prompt".to_string(),
+        }
+    }
+}
+
+fn detect_tool_permission_prompt(
+    screen_text: &str,
+    lowered: &str,
+) -> Option<ToolPermissionPromptObservation> {
+    let looks_like_prompt = lowered.contains("allow the")
+        && lowered.contains("server")
+        && lowered.contains("tool")
+        && lowered.contains("run");
+    let looks_like_tool_gate = lowered.contains("allow tool") && lowered.contains("run");
+    if !looks_like_prompt && !looks_like_tool_gate {
+        return None;
+    }
+
+    let prompt_line = screen_text
+        .lines()
+        .rev()
+        .find(|line| {
+            let lowered_line = line.to_ascii_lowercase();
+            lowered_line.contains("allow")
+                && lowered_line.contains("tool")
+                && (lowered_line.contains("run") || lowered_line.contains("server"))
+        })
+        .unwrap_or(screen_text)
+        .trim();
+
+    let tool_name = extract_quoted_value(prompt_line)
+        .or_else(|| extract_after(prompt_line, "tool ").map(|token| normalize_tool_token(&token)));
+    let server_name = extract_between(prompt_line, "the ", " server")
+        .map(|server| server.trim_end_matches(" MCP").to_string())
+        .or_else(|| {
+            tool_name
+                .as_deref()
+                .and_then(extract_server_from_qualified_tool)
+        });
+
+    Some(ToolPermissionPromptObservation {
+        server_name,
+        tool_name,
+        allow_scope: detect_tool_permission_allow_scope(lowered),
+        prompt_preview: prompt_preview(prompt_line),
+    })
+}
+
+fn detect_tool_permission_allow_scope(lowered: &str) -> ToolPermissionAllowScope {
+    let always_allow_capable = [
+        "always allow",
+        "allow always",
+        "allow this tool always",
+        "allow for all sessions",
+    ]
+    .iter()
+    .any(|needle| lowered.contains(needle));
+
+    if always_allow_capable {
+        return ToolPermissionAllowScope::SessionOrAlways;
+    }
+
+    let session_allow_capable = [
+        "allow once",
+        "allow for this session",
+        "allow this session",
+        "yes, allow",
+    ]
+    .iter()
+    .any(|needle| lowered.contains(needle));
+
+    if session_allow_capable {
+        ToolPermissionAllowScope::SessionOnly
+    } else {
+        ToolPermissionAllowScope::Unknown
+    }
+}
+
+fn extract_quoted_value(text: &str) -> Option<String> {
+    let start = text.find('"')? + 1;
+    let rest = &text[start..];
+    let end = rest.find('"')?;
+    Some(rest[..end].to_string())
+}
+
+fn extract_between(text: &str, prefix: &str, suffix: &str) -> Option<String> {
+    let start = text.find(prefix)? + prefix.len();
+    let rest = &text[start..];
+    let end = rest.find(suffix)?;
+    let value = rest[..end].trim();
+    (!value.is_empty()).then(|| value.to_string())
+}
+
+fn extract_after(text: &str, prefix: &str) -> Option<String> {
+    let start = text.to_ascii_lowercase().find(prefix)? + prefix.len();
+    let value = text[start..]
+        .split_whitespace()
+        .next()?
+        .trim_matches(|ch: char| ch == '?' || ch == ':' || ch == '"' || ch == '\'');
+    (!value.is_empty()).then(|| value.to_string())
+}
+
+fn normalize_tool_token(token: &str) -> String {
+    token
+        .trim_matches(|ch: char| ch == '?' || ch == ':' || ch == '"' || ch == '\'')
+        .to_string()
+}
+
+fn extract_server_from_qualified_tool(tool: &str) -> Option<String> {
+    let rest = tool.strip_prefix("mcp__")?;
+    let (server, _) = rest.split_once("__")?;
+    (!server.is_empty()).then(|| server.to_string())
+}
+
 fn detect_trust_prompt(lowered: &str) -> bool {
     [
         "do you trust the files in this folder",
@@ -1134,6 +1383,96 @@ mod tests {
         assert!(detect_ready_for_prompt("│ >", "│ >"));
     }
 
+    #[test]
+    fn tool_permission_prompt_blocks_worker_with_structured_event() {
+        let registry = WorkerRegistry::new();
+        let worker = registry.create("/tmp/repo-mcp", &[], true);
+
+        let blocked = registry
+            .observe(
+                &worker.worker_id,
+                "Allow the omx_memory MCP server to run tool \"project_memory_read\"?\n\
+                 1. Yes, allow once\n\
+                 2. Always allow this tool",
+            )
+            .expect("tool permission observe should succeed");
+
+        assert_eq!(blocked.status, WorkerStatus::ToolPermissionRequired);
+        assert_eq!(
+            blocked
+                .last_error
+                .as_ref()
+                .expect("tool permission error should exist")
+                .kind,
+            WorkerFailureKind::ToolPermissionGate
+        );
+        let event = blocked
+            .events
+            .iter()
+            .find(|event| event.kind == WorkerEventKind::ToolPermissionRequired)
+            .expect("tool permission event should exist");
+        assert_eq!(
+            event.payload,
+            Some(WorkerEventPayload::ToolPermissionPrompt {
+                server_name: Some("omx_memory".to_string()),
+                tool_name: Some("project_memory_read".to_string()),
+                prompt_age_seconds: 0,
+                allow_scope: ToolPermissionAllowScope::SessionOrAlways,
+                prompt_preview: prompt_preview(
+                    "Allow the omx_memory MCP server to run tool \"project_memory_read\"?",
+                ),
+            })
+        );
+
+        let readiness = registry
+            .await_ready(&worker.worker_id)
+            .expect("ready snapshot should load");
+        assert!(readiness.blocked);
+        assert!(!readiness.ready);
+    }
+
+    #[test]
+    fn startup_timeout_classifies_tool_permission_prompt() {
+        let registry = WorkerRegistry::new();
+        let worker = registry.create("/tmp/repo-mcp-timeout", &[], true);
+
+        registry
+            .observe(
+                &worker.worker_id,
+                "Allow the omx_memory MCP server to run tool \"notepad_read\"?\n\
+                 1. Yes, allow once",
+            )
+            .expect("tool permission observe should succeed");
+
+        let timed_out = registry
+            .observe_startup_timeout(&worker.worker_id, "claw prompt", true, true)
+            .expect("startup timeout observe should succeed");
+        let event = timed_out
+            .events
+            .iter()
+            .find(|event| event.kind == WorkerEventKind::StartupNoEvidence)
+            .expect("startup no evidence event should exist");
+
+        match event.payload.as_ref() {
+            Some(WorkerEventPayload::StartupNoEvidence {
+                classification,
+                evidence,
+            }) => {
+                assert_eq!(
+                    *classification,
+                    StartupFailureClassification::ToolPermissionRequired
+                );
+                assert!(evidence.tool_permission_prompt_detected);
+                assert_eq!(
+                    evidence.tool_permission_allow_scope,
+                    Some(ToolPermissionAllowScope::SessionOnly)
+                );
+                assert!(evidence.tool_permission_prompt_age_seconds.is_some());
+            }
+            _ => panic!("expected StartupNoEvidence payload"),
+        }
+    }
+
     #[test]
     fn prompt_misdelivery_is_detected_and_replay_can_be_rearmed() {
         let registry = WorkerRegistry::new();
@@ -1534,8 +1873,16 @@ mod tests {
                     "last state should be spawning"
                 );
                 assert_eq!(evidence.pane_command, "cargo test");
+                assert!(evidence.command_started_at <= evidence.pane_observed_at);
+                assert!(evidence.last_lifecycle_at <= evidence.pane_observed_at);
                 assert!(!evidence.transport_healthy);
+                assert!(!evidence.transport_health.healthy);
+                assert!(evidence
+                    .transport_health
+                    .summary
+                    .contains("transport_unhealthy"));
                 assert!(evidence.mcp_healthy);
+                assert!(evidence.mcp_health.healthy);
                 assert_eq!(*classification, StartupFailureClassification::TransportDead);
             }
             _ => panic!(
@@ -1626,16 +1973,63 @@ mod tests {
         }
     }
 
+    #[test]
+    fn startup_timeout_preserves_original_prompt_sent_timestamp() {
+        let registry = WorkerRegistry::new();
+        let worker = registry.create("/tmp/repo-prompt-timestamp", &[], true);
+
+        registry
+            .observe(&worker.worker_id, "Ready for input\n>")
+            .expect("ready observe should succeed");
+        let prompted = registry
+            .send_prompt(
+                &worker.worker_id,
+                Some("Run timestamp-sensitive work"),
+                None,
+            )
+            .expect("prompt send should succeed");
+        let sent_at = prompted
+            .prompt_sent_at
+            .expect("prompt send should record a prompt timestamp");
+
+        let timed_out = registry
+            .observe_startup_timeout(&worker.worker_id, "claw worker", true, true)
+            .expect("startup timeout observe should succeed");
+
+        let event = timed_out
+            .events
+            .iter()
+            .find(|e| e.kind == WorkerEventKind::StartupNoEvidence)
+            .expect("startup no evidence event should exist");
+
+        match event.payload.as_ref() {
+            Some(WorkerEventPayload::StartupNoEvidence { evidence, .. }) => {
+                assert_eq!(evidence.prompt_sent_at, Some(sent_at));
+                assert!(evidence.last_lifecycle_at <= evidence.pane_observed_at);
+                assert!(evidence.command_started_at <= sent_at);
+            }
+            _ => panic!("expected StartupNoEvidence payload"),
+        }
+    }
+
     #[test]
     fn startup_evidence_bundle_serializes_correctly() {
         let bundle = StartupEvidenceBundle {
             last_lifecycle_state: WorkerStatus::Running,
+            last_lifecycle_at: 1_234_567_889,
             pane_command: "test command".to_string(),
+            pane_observed_at: 1_234_567_891,
+            command_started_at: 1_234_567_800,
             prompt_sent_at: Some(1_234_567_890),
             prompt_acceptance_state: false,
             trust_prompt_detected: true,
+            tool_permission_prompt_detected: false,
+            tool_permission_prompt_age_seconds: None,
+            tool_permission_allow_scope: None,
             transport_healthy: true,
+            transport_health: StartupHealthSummary::observed("transport", true),
             mcp_healthy: false,
+            mcp_health: StartupHealthSummary::observed("mcp", false),
             elapsed_seconds: 60,
         };
 
@@ -1644,8 +2038,13 @@ mod tests {
         assert!(json.contains("\"pane_command\""));
         assert!(json.contains("\"prompt_sent_at\":1234567890"));
         assert!(json.contains("\"trust_prompt_detected\":true"));
+        assert!(json.contains("\"last_lifecycle_at\":1234567889"));
+        assert!(json.contains("\"pane_observed_at\":1234567891"));
+        assert!(json.contains("\"command_started_at\":1234567800"));
         assert!(json.contains("\"transport_healthy\":true"));
+        assert!(json.contains("\"transport_health\""));
         assert!(json.contains("\"mcp_healthy\":false"));
+        assert!(json.contains("\"mcp_health\""));
 
         let deserialized: StartupEvidenceBundle =
             serde_json::from_str(&json).expect("should deserialize");
@@ -1657,12 +2056,20 @@ mod tests {
     fn classify_startup_failure_detects_transport_dead() {
         let evidence = StartupEvidenceBundle {
             last_lifecycle_state: WorkerStatus::Spawning,
+            last_lifecycle_at: 10,
             pane_command: "test".to_string(),
+            pane_observed_at: 40,
+            command_started_at: 1,
             prompt_sent_at: None,
             prompt_acceptance_state: false,
             trust_prompt_detected: false,
+            tool_permission_prompt_detected: false,
+            tool_permission_prompt_age_seconds: None,
+            tool_permission_allow_scope: None,
             transport_healthy: false,
+            transport_health: StartupHealthSummary::observed("transport", false),
             mcp_healthy: true,
+            mcp_health: StartupHealthSummary::observed("mcp", true),
             elapsed_seconds: 30,
         };
 
@@ -1674,12 +2081,20 @@ mod tests {
     fn classify_startup_failure_defaults_to_unknown() {
         let evidence = StartupEvidenceBundle {
             last_lifecycle_state: WorkerStatus::Spawning,
+            last_lifecycle_at: 10,
             pane_command: "test".to_string(),
+            pane_observed_at: 40,
+            command_started_at: 1,
             prompt_sent_at: None,
             prompt_acceptance_state: false,
             trust_prompt_detected: false,
+            tool_permission_prompt_detected: false,
+            tool_permission_prompt_age_seconds: None,
+            tool_permission_allow_scope: None,
             transport_healthy: true,
+            transport_health: StartupHealthSummary::observed("transport", true),
             mcp_healthy: true,
+            mcp_health: StartupHealthSummary::observed("mcp", true),
             elapsed_seconds: 10,
         };
 
@@ -1687,18 +2102,54 @@ mod tests {
         assert_eq!(classification, StartupFailureClassification::Unknown);
     }
 
+    #[test]
+    fn classify_startup_failure_detects_prompt_misdelivery_after_timeout() {
+        let evidence = StartupEvidenceBundle {
+            last_lifecycle_state: WorkerStatus::ReadyForPrompt,
+            last_lifecycle_at: 10,
+            pane_command: "test".to_string(),
+            pane_observed_at: 45,
+            command_started_at: 1,
+            prompt_sent_at: Some(10),
+            prompt_acceptance_state: false,
+            trust_prompt_detected: false,
+            tool_permission_prompt_detected: false,
+            tool_permission_prompt_age_seconds: None,
+            tool_permission_allow_scope: None,
+            transport_healthy: true,
+            transport_health: StartupHealthSummary::observed("transport", true),
+            mcp_healthy: true,
+            mcp_health: StartupHealthSummary::observed("mcp", true),
+            elapsed_seconds: 31,
+        };
+
+        let classification = classify_startup_failure(&evidence);
+        assert_eq!(
+            classification,
+            StartupFailureClassification::PromptMisdelivery
+        );
+    }
+
     #[test]
     fn classify_startup_failure_detects_worker_crashed() {
         // Worker crashed scenario: transport healthy but MCP unhealthy
         // Don't have prompt in flight (no prompt_sent_at) to avoid matching PromptAcceptanceTimeout
         let evidence = StartupEvidenceBundle {
             last_lifecycle_state: WorkerStatus::Spawning,
+            last_lifecycle_at: 10,
             pane_command: "test".to_string(),
+            pane_observed_at: 40,
+            command_started_at: 1,
             prompt_sent_at: None, // No prompt sent yet
             prompt_acceptance_state: false,
             trust_prompt_detected: false,
+            tool_permission_prompt_detected: false,
+            tool_permission_prompt_age_seconds: None,
+            tool_permission_allow_scope: None,
             transport_healthy: true,
-            mcp_healthy: false, // MCP unhealthy but transport healthy suggests crash
+            transport_health: StartupHealthSummary::observed("transport", true),
+            mcp_healthy: false,
+            mcp_health: StartupHealthSummary::observed("mcp", false), // MCP unhealthy but transport healthy suggests crash
             elapsed_seconds: 45,
         };
 

rust/crates/runtime/tests/fixtures/g004_contract_bundle.valid.json

@@ -0,0 +1,81 @@
+{
+  "schemaVersion": "g004.contract.bundle.v1",
+  "laneEvents": [
+    {
+      "event": "lane.started",
+      "status": "running",
+      "emittedAt": "2026-05-14T00:00:00Z",
+      "metadata": {
+        "seq": 1,
+        "provenance": "live_lane",
+        "emitterIdentity": "worker-1",
+        "environmentLabel": "team-g004"
+      }
+    },
+    {
+      "event": "lane.finished",
+      "status": "completed",
+      "emittedAt": "2026-05-14T00:00:10Z",
+      "metadata": {
+        "seq": 2,
+        "provenance": "live_lane",
+        "emitterIdentity": "worker-1",
+        "environmentLabel": "team-g004",
+        "eventFingerprint": "terminal-fp-001"
+      }
+    }
+  ],
+  "reports": [
+    {
+      "schemaVersion": "g004.report.v1",
+      "reportId": "report-g004-fixture",
+      "identity": { "contentHash": "sha256:report-content" },
+      "projection": { "provenance": "runtime.event_projection.v1" },
+      "redaction": { "provenance": "runtime.redaction_policy.v1" },
+      "consumerCapabilities": ["facts", "field_deltas", "redaction_provenance"],
+      "findings": [
+        {
+          "kind": "fact",
+          "confidence": "high",
+          "statement": "lane event reached terminal state"
+        },
+        {
+          "kind": "hypothesis",
+          "confidence": "medium",
+          "statement": "consumer can reconcile the terminal fingerprint"
+        },
+        {
+          "kind": "negative_evidence",
+          "confidence": "high",
+          "statement": "no duplicate terminal event appears in this fixture"
+        }
+      ],
+      "fieldDeltas": [
+        {
+          "field": "/laneEvents/1/status",
+          "previousHash": "sha256:running",
+          "currentHash": "sha256:completed",
+          "attribution": "worker-1 terminal reconciliation"
+        }
+      ]
+    }
+  ],
+  "approvalTokens": [
+    {
+      "tokenId": "approval-token-fixture",
+      "owner": "leader-fixed",
+      "scope": "g004.contract.bundle.fixture",
+      "issuedAt": "2026-05-14T00:00:01Z",
+      "oneTimeUse": true,
+      "replayPreventionNonce": "nonce-fixture-001",
+      "delegationChain": [
+        {
+          "from": "leader-fixed",
+          "to": "worker-3",
+          "action": "validate-g004-contract-fixture",
+          "at": "2026-05-14T00:00:02Z"
+        }
+      ]
+    }
+  ]
+}

rust/crates/runtime/tests/fixtures/report_schema_v1/README.md

@@ -0,0 +1,11 @@
+# Report schema v1 fixture set
+
+Validated by `cargo test -p runtime report_schema -- --nocapture`.
+
+The in-code fixture in `runtime::report_schema::tests::fixture_report` covers:
+- fact / hypothesis / confidence labels
+- negative evidence with checked surfaces and query window
+- field-level delta attribution
+- canonical report id plus content hash
+- deterministic projection/redaction provenance
+- consumer capability negotiation and downgraded projections

rust/crates/runtime/tests/g004_conformance.rs

@@ -0,0 +1,80 @@
+use runtime::g004_conformance::{is_g004_contract_bundle_valid, validate_g004_contract_bundle};
+use serde_json::{json, Value};
+
+fn valid_bundle() -> Value {
+    serde_json::from_str(include_str!("fixtures/g004_contract_bundle.valid.json"))
+        .expect("valid fixture JSON should parse")
+}
+
+#[test]
+fn valid_g004_contract_bundle_fixture_passes_conformance() {
+    let fixture = valid_bundle();
+
+    let errors = validate_g004_contract_bundle(&fixture);
+
+    assert!(
+        errors.is_empty(),
+        "unexpected conformance errors: {errors:?}"
+    );
+    assert!(is_g004_contract_bundle_valid(&fixture));
+}
+
+#[test]
+fn g004_conformance_reports_machine_readable_paths_for_contract_gaps() {
+    let invalid = json!({
+        "schemaVersion": "g004.contract.bundle.v1",
+        "laneEvents": [
+            {
+                "event": "lane.finished",
+                "status": "completed",
+                "emittedAt": "2026-05-14T00:00:10Z",
+                "metadata": {
+                    "seq": 1,
+                    "provenance": "live_lane",
+                    "emitterIdentity": "worker-1",
+                    "environmentLabel": "team-g004"
+                }
+            }
+        ],
+        "reports": [
+            {
+                "schemaVersion": "g004.report.v1",
+                "reportId": "report-with-gaps",
+                "identity": { "contentHash": "sha256:report-content" },
+                "projection": { "provenance": "runtime.event_projection.v1" },
+                "redaction": { "provenance": "runtime.redaction_policy.v1" },
+                "consumerCapabilities": [],
+                "findings": [
+                    {
+                        "kind": "guess",
+                        "confidence": "certain",
+                        "statement": "bad labels should be rejected"
+                    }
+                ],
+                "fieldDeltas": []
+            }
+        ],
+        "approvalTokens": [
+            {
+                "tokenId": "approval-token-fixture",
+                "owner": "leader-fixed",
+                "scope": "g004.contract.bundle.fixture",
+                "issuedAt": "2026-05-14T00:00:01Z",
+                "oneTimeUse": false,
+                "replayPreventionNonce": "nonce-fixture-001",
+                "delegationChain": []
+            }
+        ]
+    });
+
+    let errors = validate_g004_contract_bundle(&invalid);
+    let paths: Vec<&str> = errors.iter().map(|error| error.path.as_str()).collect();
+
+    assert!(paths.contains(&"/laneEvents/0/metadata/eventFingerprint"));
+    assert!(paths.contains(&"/reports/0/consumerCapabilities"));
+    assert!(paths.contains(&"/reports/0/findings/0/kind"));
+    assert!(paths.contains(&"/reports/0/findings/0/confidence"));
+    assert!(paths.contains(&"/reports/0/fieldDeltas"));
+    assert!(paths.contains(&"/approvalTokens/0/oneTimeUse"));
+    assert!(paths.contains(&"/approvalTokens/0/delegationChain"));
+}

rust/crates/runtime/tests/integration_tests.rs

@@ -22,7 +22,7 @@ fn stale_branch_detection_flows_into_policy_engine() {
     let stale_context = LaneContext::new(
         "stale-lane",
         0,
-        Duration::from_secs(2 * 60 * 60), // 2 hours stale
+        Duration::from_hours(2), // 2 hours stale
         LaneBlocker::None,
         ReviewStatus::Pending,
         DiffScope::Full,
@@ -49,7 +49,7 @@ fn fresh_branch_does_not_trigger_stale_policy() {
     let fresh_context = LaneContext::new(
         "fresh-lane",
         0,
-        Duration::from_secs(30 * 60), // 30 min stale — under 1 hour threshold
+        Duration::from_mins(30), // 30 min stale — under 1 hour threshold
         LaneBlocker::None,
         ReviewStatus::Pending,
         DiffScope::Full,
@@ -96,9 +96,7 @@ fn green_contract_unsatisfied_blocks_merge() {
         false,
     );
 
-    // This is a conceptual test — we need a way to express "requires workspace green"
+    // The context has a test level but lacks the full green contract, so merge stays blocked.
-    // Currently LaneContext has raw green_level: u8, not a contract
-    // For now we just verify the policy condition works
     let engine = PolicyEngine::new(vec![PolicyRule::new(
         "workspace-green-required",
         PolicyCondition::GreenAt { level: 3 }, // GreenLevel::Workspace
@@ -212,8 +210,8 @@ fn end_to_end_stale_lane_gets_merge_forward_action() {
     // when: build context and evaluate policy
     let context = LaneContext::new(
         "lane-9411",
-        3,                                // Workspace green
+        3,                       // Workspace green
-        Duration::from_secs(5 * 60 * 60), // 5 hours stale, definitely over threshold
+        Duration::from_hours(5), // 5 hours stale, definitely over threshold
         LaneBlocker::None,
         ReviewStatus::Approved,
         DiffScope::Scoped,
@@ -261,13 +259,14 @@ fn end_to_end_stale_lane_gets_merge_forward_action() {
 fn fresh_approved_lane_gets_merge_action() {
     let context = LaneContext::new(
         "fresh-approved-lane",
-        3,                            // Workspace green
+        3,                       // Workspace green
-        Duration::from_secs(30 * 60), // 30 min — under 1 hour threshold = fresh
+        Duration::from_mins(30), // 30 min — under 1 hour threshold = fresh
         LaneBlocker::None,
         ReviewStatus::Approved,
         DiffScope::Scoped,
         false,
-    );
+    )
+    .with_green_contract_satisfied(true);
 
     let engine = PolicyEngine::new(vec![PolicyRule::new(
         "merge-if-green-approved-not-stale",
@@ -347,7 +346,7 @@ fn worker_provider_failure_flows_through_recovery_to_policy() {
     // (Simulating the policy check that would happen after successful recovery)
     let recovery_success = matches!(result, RecoveryResult::Recovered { .. });
     let green_level = 3; // Workspace green
-    let not_stale = Duration::from_secs(30 * 60); // 30 min — fresh
+    let not_stale = Duration::from_mins(30); // 30 min — fresh
 
     let post_recovery_context = LaneContext::new(
         "recovered-lane",
@@ -357,7 +356,8 @@ fn worker_provider_failure_flows_through_recovery_to_policy() {
         ReviewStatus::Approved,
         DiffScope::Scoped,
         false,
-    );
+    )
+    .with_green_contract_satisfied(true);
 
     let policy_engine = PolicyEngine::new(vec![
         // Rule: if recovered from failure + green + approved -> merge

rust/crates/rusty-claude-cli/tests/compact_output.rs

@@ -126,6 +126,66 @@ fn compact_flag_streaming_text_only_emits_final_message_text() {
     fs::remove_dir_all(&workspace).expect("workspace cleanup should succeed");
 }
 
+#[test]
+fn text_prompt_mode_prints_final_assistant_text_after_spinner() {
+    // given a workspace pointed at the mock Anthropic service running the
+    // streaming_text scenario which only emits a single assistant text block
+    let runtime = tokio::runtime::Runtime::new().expect("tokio runtime should build");
+    let server = runtime
+        .block_on(MockAnthropicService::spawn())
+        .expect("mock service should start");
+    let base_url = server.base_url();
+
+    let workspace = unique_temp_dir("text-prompt-mode");
+    let config_home = workspace.join("config-home");
+    let home = workspace.join("home");
+    fs::create_dir_all(&workspace).expect("workspace should exist");
+    fs::create_dir_all(&config_home).expect("config home should exist");
+    fs::create_dir_all(&home).expect("home should exist");
+
+    // when we invoke claw in normal text prompt mode for the streaming text scenario
+    let prompt = format!("{SCENARIO_PREFIX}streaming_text");
+    let output = run_claw(
+        &workspace,
+        &config_home,
+        &home,
+        &base_url,
+        &[
+            "--model",
+            "sonnet",
+            "--permission-mode",
+            "read-only",
+            &prompt,
+        ],
+    );
+
+    // then stdout should contain the final assistant text, not just spinner output
+    assert!(
+        output.status.success(),
+        "text prompt run should succeed\nstdout:\n{}\n\nstderr:\n{}",
+        String::from_utf8_lossy(&output.stdout),
+        String::from_utf8_lossy(&output.stderr),
+    );
+    let stdout = String::from_utf8(output.stdout).expect("stdout should be utf8");
+    let plain_stdout = strip_ansi_codes(&stdout);
+    assert!(
+        plain_stdout.contains("Mock streaming says hello from the parity harness."),
+        "text prompt stdout should include the assistant text ({stdout:?})"
+    );
+    assert!(
+        plain_stdout.contains("✔ ✨ Done"),
+        "text prompt stdout should still include spinner completion ({stdout:?})"
+    );
+    assert!(
+        plain_stdout
+            .lines()
+            .any(|line| line == "Mock streaming says hello from the parity harness."),
+        "text prompt stdout should print the assistant text as its own line ({stdout:?})"
+    );
+
+    fs::remove_dir_all(&workspace).expect("workspace cleanup should succeed");
+}
+
 #[test]
 fn compact_flag_with_json_output_emits_structured_json() {
     let runtime = tokio::runtime::Runtime::new().expect("tokio runtime should build");
@@ -172,7 +232,10 @@ stderr:
     );
     let stdout = String::from_utf8(output.stdout).expect("stdout should be utf8");
     let parsed: Value = serde_json::from_str(&stdout).expect("compact json stdout should parse");
-    assert_eq!(parsed["message"], "Mock streaming says hello from the parity harness.");
+    assert_eq!(
+        parsed["message"],
+        "Mock streaming says hello from the parity harness."
+    );
     assert_eq!(parsed["compact"], true);
     assert_eq!(parsed["model"], "claude-sonnet-4-6");
     assert!(parsed["usage"].is_object());
@@ -212,3 +275,21 @@ fn unique_temp_dir(label: &str) -> PathBuf {
         std::process::id()
     ))
 }
+
+fn strip_ansi_codes(input: &str) -> String {
+    let mut output = String::with_capacity(input.len());
+    let mut chars = input.chars().peekable();
+    while let Some(ch) = chars.next() {
+        if ch == '\u{1b}' && matches!(chars.peek(), Some('[')) {
+            chars.next();
+            while let Some(next) = chars.next() {
+                if ('@'..='~').contains(&next) {
+                    break;
+                }
+            }
+            continue;
+        }
+        output.push(ch);
+    }
+    output
+}

rust/crates/rusty-claude-cli/tests/compact_repl_panic.rs

@@ -0,0 +1,138 @@
+use std::fs;
+use std::io::Write;
+use std::path::PathBuf;
+use std::process::{Command, Output, Stdio};
+use std::sync::atomic::{AtomicU64, Ordering};
+use std::time::{SystemTime, UNIX_EPOCH};
+
+static TEMP_COUNTER: AtomicU64 = AtomicU64::new(0);
+
+#[test]
+fn compact_slash_command_in_repl_does_not_start_nested_tokio_runtime() {
+    // given
+    let workspace = unique_temp_dir("compact-repl-panic");
+    let config_home = workspace.join("config-home");
+    let home = workspace.join("home");
+    fs::create_dir_all(&workspace).expect("workspace should exist");
+    fs::create_dir_all(&config_home).expect("config home should exist");
+    fs::create_dir_all(&home).expect("home should exist");
+
+    // when
+    let output = run_claw_repl(&workspace, &config_home, &home, "/compact\n/exit\n");
+
+    // then
+    assert!(
+        output.status.success(),
+        "compact repl run should succeed\nstdout:\n{}\n\nstderr:\n{}",
+        String::from_utf8_lossy(&output.stdout),
+        String::from_utf8_lossy(&output.stderr),
+    );
+    let stderr = String::from_utf8(output.stderr).expect("stderr should be utf8");
+    assert!(
+        !stderr.contains("Cannot start a runtime"),
+        "stderr must not contain nested runtime panic: {stderr:?}"
+    );
+    assert!(
+        !stderr.contains("panicked at"),
+        "stderr must not contain panic output: {stderr:?}"
+    );
+
+    let stdout = String::from_utf8(output.stdout).expect("stdout should be utf8");
+    let plain_stdout = strip_ansi_codes(&stdout);
+    assert!(
+        plain_stdout.contains("Compaction skipped")
+            || plain_stdout.contains("Result           skipped")
+            || plain_stdout.contains("Result           compacted"),
+        "stdout should contain compact report output ({stdout:?})"
+    );
+
+    fs::remove_dir_all(&workspace).expect("workspace cleanup should succeed");
+}
+
+fn run_claw_repl(
+    cwd: &std::path::Path,
+    config_home: &std::path::Path,
+    home: &std::path::Path,
+    stdin: &str,
+) -> Output {
+    let mut command = python_pty_command(env!("CARGO_BIN_EXE_claw"));
+    let mut child = command
+        .current_dir(cwd)
+        .env_clear()
+        .env("ANTHROPIC_API_KEY", "test-compact-repl-key")
+        .env("CLAW_CONFIG_HOME", config_home)
+        .env("HOME", home)
+        .env("NO_COLOR", "1")
+        .env("PATH", "/usr/bin:/bin")
+        .stdin(Stdio::piped())
+        .stdout(Stdio::piped())
+        .stderr(Stdio::piped())
+        .spawn()
+        .expect("claw should launch");
+
+    child
+        .stdin
+        .as_mut()
+        .expect("stdin should be piped")
+        .write_all(stdin.as_bytes())
+        .expect("stdin should write");
+
+    child.wait_with_output().expect("claw should finish")
+}
+
+fn python_pty_command(claw: &str) -> Command {
+    let mut command = Command::new("python3");
+    command.args([
+        "-c",
+        r#"
+import os
+import pty
+import subprocess
+import sys
+
+claw = sys.argv[1]
+payload = sys.stdin.buffer.read()
+master, slave = pty.openpty()
+child = subprocess.Popen([claw], stdin=slave, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+os.close(slave)
+os.write(master, payload)
+stdout, stderr = child.communicate(timeout=30)
+os.close(master)
+sys.stdout.buffer.write(stdout)
+sys.stderr.buffer.write(stderr)
+raise SystemExit(child.returncode)
+"#,
+        claw,
+    ]);
+    command
+}
+
+fn unique_temp_dir(label: &str) -> PathBuf {
+    let millis = SystemTime::now()
+        .duration_since(UNIX_EPOCH)
+        .expect("clock should be after epoch")
+        .as_millis();
+    let counter = TEMP_COUNTER.fetch_add(1, Ordering::Relaxed);
+    std::env::temp_dir().join(format!(
+        "claw-{label}-{}-{millis}-{counter}",
+        std::process::id()
+    ))
+}
+
+fn strip_ansi_codes(input: &str) -> String {
+    let mut output = String::with_capacity(input.len());
+    let mut chars = input.chars().peekable();
+    while let Some(ch) = chars.next() {
+        if ch == '\u{1b}' && matches!(chars.peek(), Some('[')) {
+            chars.next();
+            for next in chars.by_ref() {
+                if ('@'..='~').contains(&next) {
+                    break;
+                }
+            }
+            continue;
+        }
+        output.push(ch);
+    }
+    output
+}

rust/crates/rusty-claude-cli/tests/output_format_contract.rs

@@ -22,6 +22,42 @@ fn help_emits_json_when_requested() {
         .contains("Usage:"));
 }
 
+#[test]
+fn export_help_emits_bounded_json_when_requested_384() {
+    let root = unique_temp_dir("export-help-json");
+    fs::create_dir_all(&root).expect("temp dir should exist");
+
+    let parsed = assert_json_command(&root, &["export", "--help", "--output-format", "json"]);
+    assert_eq!(parsed["kind"], "help");
+    assert_eq!(parsed["topic"], "export");
+    assert_eq!(parsed["command"], "export");
+    assert_eq!(
+        parsed["usage"],
+        "claw export [--session <id|latest>] [--output <path>] [--output-format <format>]"
+    );
+    assert_eq!(parsed["defaults"]["session"], "latest");
+    assert!(parsed["options"].as_array().expect("options").len() >= 4);
+    assert!(parsed.get("message").is_none());
+}
+
+#[test]
+fn export_help_preserves_plaintext_in_text_mode_384() {
+    let root = unique_temp_dir("export-help-text");
+    fs::create_dir_all(&root).expect("temp dir should exist");
+
+    let output = run_claw(&root, &["export", "--help"], &[]);
+    assert!(
+        output.status.success(),
+        "stdout:\n{}\n\nstderr:\n{}",
+        String::from_utf8_lossy(&output.stdout),
+        String::from_utf8_lossy(&output.stderr)
+    );
+    let stdout = String::from_utf8(output.stdout).expect("stdout utf8");
+    assert!(stdout.starts_with("Export\n"));
+    assert!(stdout.contains("Usage            claw export"));
+    serde_json::from_str::<Value>(&stdout).expect_err("text help should remain plaintext");
+}
+
 #[test]
 fn version_emits_json_when_requested() {
     let root = unique_temp_dir("version-json");
@@ -30,6 +66,15 @@ fn version_emits_json_when_requested() {
     let parsed = assert_json_command(&root, &["--output-format", "json", "version"]);
     assert_eq!(parsed["kind"], "version");
     assert_eq!(parsed["version"], env!("CARGO_PKG_VERSION"));
+    // Provenance fields must be present for binary identification (#507).
+    assert!(
+        parsed["build_date"].is_string(),
+        "build_date must be a string in version JSON"
+    );
+    assert!(
+        parsed["executable_path"].is_string(),
+        "executable_path must be a string in version JSON so callers can identify which binary is running"
+    );
 }
 
 #[test]
@@ -46,6 +91,32 @@ fn status_and_sandbox_emit_json_when_requested() {
     assert!(sandbox["filesystem_mode"].as_str().is_some());
 }
 
+#[test]
+fn status_json_surfaces_permission_mode_override_for_security_audit() {
+    let root = unique_temp_dir("status-json-permission-mode");
+    fs::create_dir_all(&root).expect("temp dir should exist");
+
+    let parsed = assert_json_command(
+        &root,
+        &[
+            "--permission-mode",
+            "read-only",
+            "--output-format",
+            "json",
+            "status",
+        ],
+    );
+
+    assert_eq!(parsed["kind"], "status");
+    assert_eq!(parsed["permission_mode"], "read-only");
+    assert!(
+        parsed["workspace"]["cwd"].as_str().is_some(),
+        "status JSON should retain workspace context with permission mode"
+    );
+
+    fs::remove_dir_all(root).expect("cleanup temp dir");
+}
+
 #[test]
 fn acp_guidance_emits_json_when_requested() {
     let root = unique_temp_dir("acp-json");
@@ -105,6 +176,18 @@ fn inventory_commands_emit_structured_json_when_requested() {
     let skills = assert_json_command(&root, &["--output-format", "json", "skills"]);
     assert_eq!(skills["kind"], "skills");
     assert_eq!(skills["action"], "list");
+
+    let plugins = assert_json_command(&root, &["--output-format", "json", "plugins"]);
+    assert_eq!(plugins["kind"], "plugin");
+    assert_eq!(plugins["action"], "list");
+    assert!(
+        plugins["reload_runtime"].is_boolean(),
+        "plugins reload_runtime should be a boolean"
+    );
+    assert!(
+        plugins["target"].is_null(),
+        "plugins target should be null when no plugin is targeted"
+    );
 }
 
 #[test]
@@ -227,7 +310,7 @@ fn doctor_and_resume_status_emit_json_when_requested() {
     assert!(summary["failures"].as_u64().is_some());
 
     let checks = doctor["checks"].as_array().expect("doctor checks");
-    assert_eq!(checks.len(), 6);
+    assert_eq!(checks.len(), 7);
     let check_names = checks
         .iter()
         .map(|check| {
@@ -244,6 +327,7 @@ fn doctor_and_resume_status_emit_json_when_requested() {
             "config",
             "install source",
             "workspace",
+            "boot preflight",
             "sandbox",
             "system"
         ]
@@ -269,6 +353,14 @@ fn doctor_and_resume_status_emit_json_when_requested() {
     assert!(workspace["cwd"].as_str().is_some());
     assert!(workspace["in_git_repo"].is_boolean());
 
+    let boot_preflight = checks
+        .iter()
+        .find(|check| check["name"] == "boot preflight")
+        .expect("boot preflight check");
+    assert!(boot_preflight["boot_preflight"]["repo"]["exists"].is_boolean());
+    assert!(boot_preflight["boot_preflight"]["mcp_startup"]["eligible"].is_boolean());
+    assert!(boot_preflight["boot_preflight"]["required_binaries"].is_array());
+
     let sandbox = checks
         .iter()
         .find(|check| check["name"] == "sandbox")
@@ -348,6 +440,62 @@ fn resumed_inventory_commands_emit_structured_json_when_requested() {
     assert_eq!(skills["action"], "list");
     assert!(skills["summary"]["total"].is_number());
     assert!(skills["skills"].is_array());
+
+    let agents = assert_json_command_with_env(
+        &root,
+        &[
+            "--output-format",
+            "json",
+            "--resume",
+            session_path.to_str().expect("utf8 session path"),
+            "/agents",
+        ],
+        &[
+            (
+                "CLAW_CONFIG_HOME",
+                config_home.to_str().expect("utf8 config home"),
+            ),
+            ("HOME", home.to_str().expect("utf8 home")),
+        ],
+    );
+    assert_eq!(agents["kind"], "agents");
+    assert_eq!(agents["action"], "list");
+    assert!(
+        agents["agents"].is_array(),
+        "agents field must be a JSON array"
+    );
+    assert!(
+        agents["count"].is_number(),
+        "count must be a number, not a text render"
+    );
+
+    let plugins = assert_json_command_with_env(
+        &root,
+        &[
+            "--output-format",
+            "json",
+            "--resume",
+            session_path.to_str().expect("utf8 session path"),
+            "/plugins",
+        ],
+        &[
+            (
+                "CLAW_CONFIG_HOME",
+                config_home.to_str().expect("utf8 config home"),
+            ),
+            ("HOME", home.to_str().expect("utf8 home")),
+        ],
+    );
+    assert_eq!(plugins["kind"], "plugin");
+    assert_eq!(plugins["action"], "list");
+    assert!(
+        plugins["reload_runtime"].is_boolean(),
+        "plugins reload_runtime should be a boolean"
+    );
+    assert!(
+        plugins["target"].is_null(),
+        "plugins target should be null when no plugin is targeted"
+    );
 }
 
 #[test]
@@ -384,6 +532,44 @@ fn resumed_version_and_init_emit_structured_json_when_requested() {
     assert!(root.join("CLAUDE.md").exists());
 }
 
+#[test]
+fn config_section_json_emits_section_and_value() {
+    let root = unique_temp_dir("config-section-json");
+    fs::create_dir_all(&root).expect("temp dir should exist");
+
+    // Without a section: should return base envelope (no section field).
+    let base = assert_json_command(&root, &["--output-format", "json", "config"]);
+    assert_eq!(base["kind"], "config");
+    assert!(base["loaded_files"].is_number());
+    assert!(base["merged_keys"].is_number());
+    assert!(
+        base.get("section").is_none(),
+        "no section field without section arg"
+    );
+
+    // With a known section: should add section + section_value fields.
+    for section in &["model", "env", "hooks", "plugins"] {
+        let result = assert_json_command(&root, &["--output-format", "json", "config", section]);
+        assert_eq!(result["kind"], "config", "section={section}");
+        assert_eq!(
+            result["section"].as_str(),
+            Some(*section),
+            "section field must match requested section, got {result:?}"
+        );
+        assert!(
+            result.get("section_value").is_some(),
+            "section_value field must be present for section={section}"
+        );
+    }
+
+    // With an unsupported section: should return ok:false + error field.
+    let bad = assert_json_command(&root, &["--output-format", "json", "config", "unknown"]);
+    assert_eq!(bad["kind"], "config");
+    assert_eq!(bad["ok"], false);
+    assert!(bad["error"].as_str().is_some());
+    assert!(bad["section"].as_str().is_some());
+}
+
 fn assert_json_command(current_dir: &Path, args: &[&str]) -> Value {
     assert_json_command_with_env(current_dir, args, &[])
 }

rust/crates/rusty-claude-cli/tests/resume_slash_commands.rs

@@ -227,6 +227,8 @@ fn resumed_status_command_emits_structured_json_when_requested() {
     // given
     let temp_dir = unique_temp_dir("resume-status-json");
     fs::create_dir_all(&temp_dir).expect("temp dir should exist");
+    let config_home = temp_dir.join("config-home");
+    fs::create_dir_all(&config_home).expect("isolated config home should exist");
     let session_path = temp_dir.join("session.jsonl");
 
     let mut session = workspace_session(&temp_dir);
@@ -238,7 +240,9 @@ fn resumed_status_command_emits_structured_json_when_requested() {
         .expect("session should persist");
 
     // when
-    let output = run_claw(
+    // Use an isolated CLAW_CONFIG_HOME so ~/.claw/settings.json is not loaded,
+    // which would cause loaded_config_files to be non-zero (#65).
+    let output = run_claw_with_env(
         &temp_dir,
         &[
             "--output-format",
@@ -247,6 +251,7 @@ fn resumed_status_command_emits_structured_json_when_requested() {
             session_path.to_str().expect("utf8 path"),
             "/status",
         ],
+        &[("CLAW_CONFIG_HOME", config_home.to_str().expect("utf8 path"))],
     );
 
     // then

rust/crates/tools/src/lane_completion.rs

@@ -56,6 +56,7 @@ pub(crate) fn detect_lane_completion(
     Some(LaneContext {
         lane_id: output.agent_id.clone(),
         green_level: 3, // Workspace green
+        green_contract_satisfied: true,
         branch_freshness: std::time::Duration::from_secs(0),
         blocker: LaneBlocker::None,
         review_status: ReviewStatus::Approved,
@@ -165,6 +166,7 @@ mod tests {
         let context = LaneContext {
             lane_id: "completed-lane".to_string(),
             green_level: 3,
+            green_contract_satisfied: true,
             branch_freshness: std::time::Duration::from_secs(0),
             blocker: LaneBlocker::None,
             review_status: ReviewStatus::Approved,

rust/crates/tools/tests/path_scope_enforcement.rs

@@ -0,0 +1,205 @@
+use runtime::{permission_enforcer::PermissionEnforcer, PermissionMode, PermissionPolicy};
+use serde_json::json;
+use std::fs;
+use std::path::{Path, PathBuf};
+use std::sync::{Mutex, OnceLock};
+use tools::{mvp_tool_specs, GlobalToolRegistry};
+
+fn env_lock() -> &'static Mutex<()> {
+    static LOCK: OnceLock<Mutex<()>> = OnceLock::new();
+    LOCK.get_or_init(|| Mutex::new(()))
+}
+
+fn temp_path(name: &str) -> PathBuf {
+    let unique = std::time::SystemTime::now()
+        .duration_since(std::time::UNIX_EPOCH)
+        .expect("time")
+        .as_nanos();
+    std::env::temp_dir().join(format!("claw-path-scope-{unique}-{name}"))
+}
+
+fn workspace_write_registry() -> GlobalToolRegistry {
+    let policy = mvp_tool_specs().into_iter().fold(
+        PermissionPolicy::new(PermissionMode::WorkspaceWrite),
+        |policy, spec| policy.with_tool_requirement(spec.name, spec.required_permission),
+    );
+    GlobalToolRegistry::builtin().with_enforcer(PermissionEnforcer::new(policy))
+}
+
+fn run_bash(command: &str) -> Result<String, String> {
+    workspace_write_registry().execute("bash", &json!({ "command": command }))
+}
+
+fn run_powershell(command: &str) -> Result<String, String> {
+    workspace_write_registry().execute("PowerShell", &json!({ "command": command }))
+}
+
+fn run_read_file(path: &Path) -> Result<String, String> {
+    workspace_write_registry().execute("read_file", &json!({ "path": path.display().to_string() }))
+}
+
+fn assert_permission_denied(result: Result<String, String>, case_name: &str) {
+    let err = result
+        .unwrap_err_or_else(|ok| panic!("{case_name} should be denied before execution, got {ok}"));
+    assert!(
+        (err.contains("requires danger-full-access permission")
+            || err.contains("requires \'danger-full-access\' permission"))
+            || err.contains("current mode is workspace-write")
+            || err.contains("escapes workspace"),
+        "{case_name} should fail in permission enforcement, got: {err}"
+    );
+}
+
+trait UnwrapErrOrElse<T, E> {
+    fn unwrap_err_or_else<F: FnOnce(T) -> E>(self, op: F) -> E;
+}
+
+impl<T, E> UnwrapErrOrElse<T, E> for Result<T, E> {
+    fn unwrap_err_or_else<F: FnOnce(T) -> E>(self, op: F) -> E {
+        match self {
+            Ok(value) => op(value),
+            Err(error) => error,
+        }
+    }
+}
+
+fn with_cwd<T>(cwd: &Path, f: impl FnOnce() -> T) -> T {
+    let previous = std::env::current_dir().expect("current dir");
+    std::env::set_current_dir(cwd).expect("set cwd");
+    let result = f();
+    std::env::set_current_dir(previous).expect("restore cwd");
+    result
+}
+
+#[test]
+fn direct_paths_allow_workspace_file_and_deny_absolute_outside_file() {
+    let _guard = env_lock()
+        .lock()
+        .unwrap_or_else(std::sync::PoisonError::into_inner);
+    let root = temp_path("direct");
+    fs::create_dir_all(root.join("src")).expect("create workspace");
+    fs::write(root.join("src/lib.rs"), "workspace\n").expect("write workspace file");
+    let outside = temp_path("direct-outside.txt");
+    fs::write(&outside, "secret\n").expect("write outside file");
+
+    with_cwd(&root, || {
+        let allowed = run_bash("cat src/lib.rs").expect("workspace-relative read should execute");
+        assert!(allowed.contains("workspace"));
+        assert_permission_denied(
+            run_bash(&format!("cat {}", outside.display())),
+            "absolute outside file",
+        );
+    });
+
+    let _ = fs::remove_dir_all(root);
+    let _ = fs::remove_file(outside);
+}
+
+#[test]
+fn file_tool_direct_outside_path_is_denied_before_reading() {
+    let _guard = env_lock()
+        .lock()
+        .unwrap_or_else(std::sync::PoisonError::into_inner);
+    let root = temp_path("file-tool-direct");
+    fs::create_dir_all(&root).expect("create workspace");
+    let outside = temp_path("file-tool-secret.txt");
+    fs::write(&outside, "secret\n").expect("write outside file");
+
+    with_cwd(&root, || {
+        assert_permission_denied(run_read_file(&outside), "read_file outside workspace");
+    });
+
+    let _ = fs::remove_dir_all(root);
+    let _ = fs::remove_file(outside);
+}
+
+#[cfg(unix)]
+#[test]
+fn symlink_resolving_outside_workspace_is_denied_before_execution() {
+    let _guard = env_lock()
+        .lock()
+        .unwrap_or_else(std::sync::PoisonError::into_inner);
+    let root = temp_path("symlink");
+    fs::create_dir_all(&root).expect("create workspace");
+    let outside = temp_path("symlink-secret.txt");
+    fs::write(&outside, "secret\n").expect("write outside file");
+    std::os::unix::fs::symlink(&outside, root.join("secret-link")).expect("create symlink");
+
+    with_cwd(&root, || {
+        assert_permission_denied(run_bash("cat secret-link"), "outside symlink");
+    });
+
+    let _ = fs::remove_dir_all(root);
+    let _ = fs::remove_file(outside);
+}
+
+#[test]
+fn shell_expansion_and_glob_parent_traversal_are_denied_before_execution() {
+    let _guard = env_lock()
+        .lock()
+        .unwrap_or_else(std::sync::PoisonError::into_inner);
+    let root = temp_path("expansion");
+    fs::create_dir_all(&root).expect("create workspace");
+
+    with_cwd(&root, || {
+        for (name, command) in [
+            ("parent glob", "ls ../*"),
+            ("PWD parent expansion", "cat $PWD/../secret.txt"),
+            ("braced PWD parent expansion", "cat ${PWD}/../secret.txt"),
+            (
+                "command substitution parent expansion",
+                "cat $(pwd)/../secret.txt",
+            ),
+        ] {
+            assert_permission_denied(run_bash(command), name);
+        }
+    });
+
+    let _ = fs::remove_dir_all(root);
+}
+
+#[test]
+fn nested_worktree_paths_are_allowed_but_parent_escape_is_denied() {
+    let _guard = env_lock()
+        .lock()
+        .unwrap_or_else(std::sync::PoisonError::into_inner);
+    let root = temp_path("worktree");
+    let worktree = root.join("main").join("linked-worktree");
+    fs::create_dir_all(worktree.join("src")).expect("create worktree");
+    fs::write(worktree.join("src/lib.rs"), "worktree\n").expect("write worktree file");
+
+    with_cwd(&worktree, || {
+        let allowed =
+            run_bash("cat src/lib.rs").expect("nested worktree-relative read should execute");
+        assert!(allowed.contains("worktree"));
+        assert_permission_denied(run_bash("cat ../../outside.txt"), "worktree parent escape");
+    });
+
+    let _ = fs::remove_dir_all(root);
+}
+
+#[test]
+fn windows_style_absolute_paths_are_denied_before_execution() {
+    for (name, command) in [
+        (
+            "windows drive backslash",
+            r"cat C:\Users\attacker\secret.txt",
+        ),
+        ("windows drive slash", r"cat C:/Users/attacker/secret.txt"),
+    ] {
+        assert_permission_denied(run_bash(command), name);
+    }
+
+    for (name, command) in [
+        (
+            "powershell windows drive backslash",
+            r"Get-Content -Path C:\Users\attacker\secret.txt",
+        ),
+        (
+            "powershell windows drive slash",
+            r"Get-Content -Path C:/Users/attacker/secret.txt",
+        ),
+    ] {
+        assert_permission_denied(run_powershell(command), name);
+    }
+}

scripts/cc2_board.py

@@ -0,0 +1,54 @@
+#!/usr/bin/env python3
+"""Canonical CC2 board command wrapper.
+
+This script intentionally delegates to the richer G001 board generator,
+validator, and Markdown renderer so all entrypoints enforce the same schema.
+"""
+from __future__ import annotations
+
+import argparse
+import subprocess
+import sys
+from pathlib import Path
+
+
+def run(cmd: list[str], cwd: Path) -> int:
+    return subprocess.run(cmd, cwd=str(cwd)).returncode
+
+
+def main(argv: list[str] | None = None) -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("command", choices=["generate", "validate"])
+    parser.add_argument("--repo-root", type=Path, default=Path.cwd(), help="repository root containing ROADMAP.md")
+    parser.add_argument("--context-root", type=Path, default=None, help="accepted for compatibility; source .omx is auto-detected by the generator")
+    parser.add_argument("--board-json", default=".omx/cc2/board.json")
+    parser.add_argument("--board-md", default=".omx/cc2/board.md")
+    args = parser.parse_args(argv)
+
+    repo_root = args.repo_root.resolve()
+    board_json = repo_root / args.board_json
+    board_md = repo_root / args.board_md
+    generator = repo_root / "scripts" / "generate_cc2_board.py"
+    validator = repo_root / "scripts" / "validate_cc2_board.py"
+    renderer = repo_root / ".omx" / "cc2" / "render_board_md.py"
+
+    if args.command == "generate":
+        rc = run([sys.executable, str(generator), "--repo-root", str(repo_root), "--out-dir", str(board_json.parent)], repo_root)
+        if rc:
+            return rc
+        return run([sys.executable, str(renderer), str(board_json), str(board_md)], repo_root)
+
+    checks = [
+        [sys.executable, str(validator), "--repo-root", str(repo_root), "--board", str(board_json)],
+        [sys.executable, str(renderer), str(board_json), str(board_md), "--check"],
+    ]
+    for cmd in checks:
+        rc = run(cmd, repo_root)
+        if rc:
+            return rc
+    print(f"CC2 board validation PASS: {board_json} and {board_md} are canonical and in sync")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

scripts/dogfood-build.sh

@@ -0,0 +1,68 @@
+#!/usr/bin/env bash
+# dogfood-build.sh — Build claw from current checkout and verify provenance.
+#
+# Injects GIT_SHA at build time so version JSON is non-null.
+# Suppresses Cargo compile noise on stderr.
+# Prints the verified binary path on success. Use as:
+#
+#   CLAW=$(bash scripts/dogfood-build.sh)
+#
+# Then dogfood with config isolation (avoids real user config bleeding in):
+#
+#   CLAW_CONFIG_HOME=$(mktemp -d) $CLAW plugins list --output-format json
+#
+set -euo pipefail
+
+REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
+RUST_DIR="$REPO_ROOT/rust"
+BINARY="$RUST_DIR/target/debug/claw"
+EXPECTED_SHA="$(git -C "$REPO_ROOT" rev-parse --short HEAD)"
+
+echo "▶ Building claw from $REPO_ROOT" >&2
+echo "  Commit: $(git -C "$REPO_ROOT" log --oneline -1)" >&2
+
+# Inject GIT_SHA so version JSON returns a non-null sha.
+# Redirect cargo stderr to /dev/null to suppress compile noise;
+# on build failure cargo exits non-zero and set -e aborts.
+if ! GIT_SHA="$EXPECTED_SHA" cargo build \
+        --manifest-path "$RUST_DIR/Cargo.toml" \
+        -p rusty-claude-cli -q 2>/dev/null; then
+    # Re-run with visible output so the user sees the error
+    echo "✗ Build failed — rerunning with output:" >&2
+    GIT_SHA="$EXPECTED_SHA" cargo build \
+        --manifest-path "$RUST_DIR/Cargo.toml" \
+        -p rusty-claude-cli 2>&1 | sed 's/^/  /' >&2
+    exit 1
+fi
+
+if [[ ! -x "$BINARY" ]]; then
+    echo "✗ Binary not found at $BINARY" >&2
+    exit 1
+fi
+
+BINARY_SHA=$("$BINARY" version --output-format json 2>/dev/null \
+    | python3 -c "import sys,json; d=json.load(sys.stdin); print(d.get('git_sha') or 'null')" 2>/dev/null \
+    || echo "null")
+
+if [[ "$BINARY_SHA" == "null" || -z "$BINARY_SHA" ]]; then
+    echo "✗ Provenance check failed: binary reports git_sha: null" >&2
+    exit 1
+fi
+
+if [[ "$BINARY_SHA" != "$EXPECTED_SHA" ]]; then
+    echo "✗ Provenance mismatch: binary=$BINARY_SHA, HEAD=$EXPECTED_SHA" >&2
+    exit 1
+fi
+
+echo "✓ Binary verified: $BINARY_SHA == HEAD" >&2
+echo "" >&2
+echo "  export CLAW=$BINARY" >&2
+echo "" >&2
+echo "  Dogfood with isolated config (no real user config on stderr):" >&2
+echo "    CLAW_ISOLATED=\$(mktemp -d)" >&2
+echo "    CLAW_CONFIG_HOME=\$CLAW_ISOLATED \$CLAW plugins list --output-format json" >&2
+echo "    rm -rf \$CLAW_ISOLATED" >&2
+echo "" >&2
+echo "  cargo run overhead: ~1s/invocation vs 7ms for pre-built binary." >&2
+echo "  Prefer pre-built binary (\$CLAW) for dogfood loops." >&2
+echo "$BINARY"

scripts/fmt.sh

@@ -0,0 +1,7 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
+cd "$REPO_ROOT/rust"
+exec cargo fmt "$@"

scripts/generate_cc2_board.py

@@ -0,0 +1,525 @@
+#!/usr/bin/env python3
+"""Generate the canonical Claw Code 2.0 execution board from frozen roadmap evidence."""
+from __future__ import annotations
+
+import argparse
+import hashlib
+import json
+import re
+import subprocess
+import sys
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Any
+
+REQUIRED_ITEM_FIELDS = [
+    "id",
+    "title",
+    "source_anchor",
+    "source_type",
+    "release_bucket",
+    "status",
+    "dependencies",
+    "verification_required",
+    "deferral_rationale",
+]
+STATUSES = {
+    "context",
+    "active",
+    "open",
+    "done_verify",
+    "stale_done",
+    "superseded",
+    "deferred_with_rationale",
+    "rejected_not_claw",
+}
+RELEASE_BUCKETS = {
+    "alpha_blocker",
+    "beta_adoption",
+    "ga_ecosystem",
+    "post_2_0_research",
+    "rejected_not_claw",
+    "context",
+    "2.x_intake",
+}
+
+STRUCTURAL_HEADINGS = {
+    "Clawable Coding Harness Roadmap",
+    "Goal",
+    'Definition of "clawable"',
+    "Current Pain Points",
+    "Product Principles",
+    "Roadmap",
+    "Immediate Backlog (from current real pain)",
+    "Deployment Architecture Gap (filed from dogfood 2026-04-08)",
+    "Startup Friction Gap: No Default trusted_roots in Settings (filed 2026-04-08)",
+    "Observability Transport Decision (filed 2026-04-08)",
+    "Provider Routing: Model-Name Prefix Must Win Over Env-Var Presence (fixed 2026-04-08, `0530c50`)",
+}
+
+CATEGORY_KEYWORDS = [
+    ("security", ["security", "sandbox", "permission", "trust", "approval-token", "denied"]),
+    ("windows_install", ["windows", "install", "path", "release", "binary", "container"]),
+    ("provider", ["provider", "model", "openai", "anthropic", "ollama", "llama", "vllm", "credential"]),
+    ("sessions", ["session", "resume", "compact", "context-window", "thread"]),
+    ("docs_license", ["docs", "readme", "usage", "license", "help", "onboarding"]),
+    ("ide_acp", ["zed", "acp", "editor", "daemon"]),
+    ("plugin_mcp", ["plugin", "mcp", "marketplace", "server"]),
+    ("event_report", ["event", "report", "schema", "projection", "redaction", "clawhip", "lane"]),
+    ("branch_recovery", ["branch", "stale", "recovery", "green", "flake"]),
+    ("boot", ["boot", "worker", "startup", "ready", "prompt"]),
+    ("task_policy", ["task", "policy", "claw-native", "dashboard", "lane board"]),
+    ("ux_tui", ["tui", "statusline", "keymap", "clickable", "copy", "paste"]),
+    ("anti_slop", ["spam", "slop", "issue hygiene", "bot"]),
+]
+
+@dataclass(frozen=True)
+class RoadmapRecord:
+    line: int
+    level: int
+    title: str
+    path: str
+    source_type: str
+    ordinal: int | None = None
+
+
+def sha256_prefix(path: Path, length: int = 16) -> str:
+    return hashlib.sha256(path.read_bytes()).hexdigest()[:length]
+
+
+def slugify(text: str, limit: int = 54) -> str:
+    slug = re.sub(r"[^a-z0-9]+", "-", text.lower()).strip("-")
+    return slug[:limit].strip("-") or "item"
+
+
+def find_source_omx(repo_root: Path) -> Path:
+    candidates = []
+    env = None
+    try:
+        import os
+        env = os.environ.get("CC2_SOURCE_OMX")
+    except Exception:
+        env = None
+    if env:
+        candidates.append(Path(env).expanduser())
+    candidates.append(repo_root / ".omx")
+    candidates.extend(parent / ".omx" for parent in repo_root.parents)
+    for candidate in candidates:
+        if (candidate / "plans" / "claw-code-2-0-adaptive-plan.md").exists() and (candidate / "research").exists():
+            return candidate
+    raise FileNotFoundError("could not locate source .omx with plans/claw-code-2-0-adaptive-plan.md and research/")
+
+
+def parse_roadmap(path: Path) -> tuple[list[RoadmapRecord], list[RoadmapRecord]]:
+    headings: list[RoadmapRecord] = []
+    actions: list[RoadmapRecord] = []
+    stack: list[tuple[str, int, int]] = []
+    for line_no, line in enumerate(path.read_text(encoding="utf-8").splitlines(), 1):
+        heading = re.match(r"^(#{1,6})\s+(.*?)(?:\s+#+)?\s*$", line)
+        if heading:
+            level = len(heading.group(1))
+            title = heading.group(2).strip()
+            stack = [entry for entry in stack if entry[1] < level] + [(title, level, line_no)]
+            headings.append(RoadmapRecord(line_no, level, title, " > ".join(entry[0] for entry in stack), "roadmap_heading"))
+            continue
+        ordered = re.match(r"^(\s*)(\d+)\.\s+(.+?)\s*$", line)
+        if ordered and len(ordered.group(1)) <= 4:
+            title = ordered.group(3).strip()
+            if len(title) > 10:
+                actions.append(
+                    RoadmapRecord(
+                        line_no,
+                        len(stack[-1][0]) if stack else 0,
+                        title,
+                        " > ".join(entry[0] for entry in stack),
+                        "roadmap_action",
+                        int(ordered.group(2)),
+                    )
+                )
+    return headings, actions
+
+
+def category_for(text: str) -> str:
+    lower = text.lower()
+    for category, needles in CATEGORY_KEYWORDS:
+        if any(needle in lower for needle in needles):
+            return category
+    return "governance"
+
+
+def stream_for(record: RoadmapRecord) -> str:
+    title = record.title.lower()
+    path = record.path.lower()
+    combined = f"{path} {title}"
+    if "phase 1" in combined or category_for(combined) == "boot":
+        return "stream_1_worker_boot_session_control"
+    if "phase 2" in combined or category_for(combined) == "event_report":
+        return "stream_2_event_reporting_contracts"
+    if "phase 3" in combined or category_for(combined) == "branch_recovery":
+        return "stream_3_branch_test_recovery"
+    if "phase 4" in combined or category_for(combined) == "task_policy":
+        return "stream_4_claws_first_execution"
+    if "phase 5" in combined or category_for(combined) == "plugin_mcp":
+        return "stream_5_plugin_mcp_lifecycle"
+    if any(k in combined for k in ["windows", "install", "provider", "docs", "license", "session hygiene", "compact"]):
+        return "adoption_overlay"
+    if any(k in combined for k in ["zed", "acp", "desktop", "marketplace", "package"]):
+        return "parity_overlay"
+    return "stream_0_governance"
+
+
+def release_bucket_for(record: RoadmapRecord, status: str) -> str:
+    combined = f"{record.path} {record.title}".lower()
+    category = category_for(combined)
+    if status == "context":
+        return "context"
+    if status == "rejected_not_claw":
+        return "rejected_not_claw"
+    if any(k in combined for k in ["phase 1", "phase 2", "phase 3", "phase 4", "p0", "p1", "security", "sandbox", "trust", "worker", "event", "branch freshness"]):
+        return "alpha_blocker"
+    if category in {"windows_install", "provider", "sessions", "docs_license", "anti_slop"}:
+        return "beta_adoption"
+    if category in {"plugin_mcp", "ide_acp", "ux_tui"}:
+        return "ga_ecosystem"
+    if any(k in combined for k in ["desktop", "share", "cloud", "research", "post-2.0", "future"]):
+        return "post_2_0_research"
+    if "pinpoint" in combined:
+        return "alpha_blocker"
+    return "beta_adoption"
+
+
+def status_for(record: RoadmapRecord) -> str:
+    title = record.title
+    combined = f"{record.path} {title}".lower()
+    if record.source_type == "roadmap_heading" and (record.level <= 2 or title in STRUCTURAL_HEADINGS):
+        # Phase headings are active work containers; other h1/h2 prose headings are context unless fixed/deferred wording says otherwise.
+        if title.startswith("Phase "):
+            return "active"
+        if "pinpoint" not in title.lower() and not any(word in combined for word in ["gap", "routing"]):
+            return "context"
+    if any(word in combined for word in ["rejected_not_claw", "not claw", "outside claw"]):
+        return "rejected_not_claw"
+    if "superseded" in combined:
+        return "superseded"
+    if "deferred" in combined or "post-2.0" in combined or "post_2_0" in combined:
+        return "deferred_with_rationale"
+    if any(word in combined for word in ["done", "implemented", "fixed", "verified", "re-verified", "landed", "green"]):
+        if any(word in combined for word in ["stale", "old filing", "original filing below", "no longer reproduces"]):
+            return "stale_done"
+        return "done_verify"
+    if title.lower().startswith(("evidence for", "trace path", "actual root cause", "meta-lesson")):
+        return "context"
+    return "open" if "pinpoint" in combined or record.source_type == "roadmap_action" else "active"
+
+
+def deferral_for(record: RoadmapRecord, status: str) -> str:
+    if status == "deferred_with_rationale":
+        return "Deferred by roadmap/approved plan until prerequisite contracts or post-2.0 research admission gates are satisfied."
+    if status == "rejected_not_claw":
+        return "Rejected because the source describes clone-only breadth or behavior outside Claw's machine-truth/clawable-harness identity."
+    if status == "superseded":
+        return "Superseded by a newer roadmap entry or canonical Rust/control-plane contract; keep only for audit traceability."
+    if status == "stale_done":
+        return "Marked done in roadmap but needs freshness re-verification before being used as release evidence."
+    return ""
+
+
+def verification_for(record: RoadmapRecord, status: str) -> str:
+    if status == "context":
+        return "none_context_only"
+    if status in {"done_verify", "stale_done"}:
+        return "verify_existing_evidence_and_regression_guard"
+    cat = category_for(f"{record.path} {record.title}")
+    if cat == "docs_license":
+        return "docs_snapshot_or_help_output_check"
+    if cat == "windows_install":
+        return "install_matrix_or_cross_platform_smoke"
+    if cat == "provider":
+        return "provider_routing_contract_test"
+    if cat == "plugin_mcp":
+        return "plugin_mcp_lifecycle_contract_test"
+    if cat == "event_report":
+        return "schema_golden_fixture_or_consumer_contract_test"
+    if cat == "branch_recovery":
+        return "git_fixture_or_recovery_recipe_test"
+    if cat == "boot":
+        return "worker_boot_state_machine_or_cli_json_contract_test"
+    return "targeted_regression_or_acceptance_test_required"
+
+
+def dependencies_for(record: RoadmapRecord, status: str) -> list[str]:
+    combined = f"{record.path} {record.title}".lower()
+    deps: list[str] = []
+    if status == "context":
+        return deps
+    if "phase 2" in combined or category_for(combined) == "event_report":
+        deps.append("stream_1_worker_boot_session_control")
+    if "phase 3" in combined or category_for(combined) == "branch_recovery":
+        deps.append("stream_2_event_reporting_contracts")
+    if "phase 4" in combined or category_for(combined) == "task_policy":
+        deps.append("stream_2_event_reporting_contracts")
+    if "phase 5" in combined or category_for(combined) == "plugin_mcp":
+        deps.append("stream_1_worker_boot_session_control")
+    if any(k in combined for k in ["zed", "acp", "desktop", "marketplace"]):
+        deps.append("stable_alpha_contracts")
+    if any(k in combined for k in ["provider", "install", "windows", "docs", "license"]):
+        deps.append("adoption_overlay_triage")
+    return sorted(set(deps))
+
+
+def roadmap_item(record: RoadmapRecord, index: int) -> dict[str, Any]:
+    status = status_for(record)
+    item_id = f"CC2-RM-{'H' if record.source_type == 'roadmap_heading' else 'A'}{index:04d}-{slugify(record.title, 40)}"
+    bucket = release_bucket_for(record, status)
+    return {
+        "id": item_id,
+        "title": record.title,
+        "source_anchor": f"ROADMAP.md:L{record.line}",
+        "source_type": record.source_type,
+        "source_path": "ROADMAP.md",
+        "source_context": record.path,
+        "source_line": record.line,
+        "source_level": record.level if record.source_type == "roadmap_heading" else None,
+        "source_ordinal": record.ordinal,
+        "release_bucket": bucket,
+        "lifecycle_status": status,
+        "status": status,
+        "category": category_for(f"{record.path} {record.title}"),
+        "owner_lane": stream_for(record),
+        "dependencies": dependencies_for(record, status),
+        "verification_required": verification_for(record, status),
+        "deferral_rationale": deferral_for(record, status),
+    }
+
+
+def load_json(path: Path) -> Any:
+    return json.loads(path.read_text(encoding="utf-8"))
+
+
+def issue_item(issue: dict[str, Any], source_name: str, source_type: str, bucket: str) -> dict[str, Any]:
+    title = issue.get("title") or f"Issue #{issue.get('number')}"
+    number = issue.get("number")
+    body = f"{title} {issue.get('body') or ''}"
+    status = "open" if issue.get("state", "OPEN").lower() != "closed" else "done_verify"
+    return {
+        "id": f"CC2-ISSUE-{source_name.upper()}-{number}",
+        "title": title,
+        "source_anchor": f".omx/research/{source_name}.json#issue-{number}",
+        "source_type": source_type,
+        "source_path": f".omx/research/{source_name}.json",
+        "issue_number": number,
+        "issue_url": issue.get("url"),
+        "release_bucket": bucket,
+        "lifecycle_status": status,
+        "status": status,
+        "category": category_for(body),
+        "owner_lane": stream_for(RoadmapRecord(0, 0, title, title, source_type)),
+        "dependencies": ["roadmap_board_triage"],
+        "verification_required": "issue_acceptance_repro_or_triage_decision",
+        "deferral_rationale": "Latest issue intake is admitted only when it matches freeze/admission rules; otherwise remains 2.x_intake." if bucket == "2.x_intake" else "",
+    }
+
+
+def repo_context_item(meta: dict[str, Any], source_name: str) -> dict[str, Any]:
+    owner = meta.get("nameWithOwner", source_name)
+    return {
+        "id": f"CC2-PARITY-{source_name.upper()}-REPO-CONTEXT",
+        "title": f"Parity source metadata: {owner}",
+        "source_anchor": f".omx/research/{source_name}-repo.json",
+        "source_type": "parity_repo_context",
+        "source_path": f".omx/research/{source_name}-repo.json",
+        "release_bucket": "context",
+        "lifecycle_status": "context",
+        "status": "context",
+        "category": "governance",
+        "owner_lane": "parity_overlay",
+        "dependencies": [],
+        "verification_required": "none_context_only",
+        "deferral_rationale": "",
+        "repo": {
+            "nameWithOwner": owner,
+            "url": meta.get("url"),
+            "pushedAt": meta.get("pushedAt"),
+            "latestRelease": meta.get("latestRelease"),
+            "licenseInfo": meta.get("licenseInfo"),
+        },
+    }
+
+
+def summarize_counts(items: list[dict[str, Any]], key: str) -> dict[str, int]:
+    out: dict[str, int] = {}
+    for item in items:
+        out[item[key]] = out.get(item[key], 0) + 1
+    return dict(sorted(out.items()))
+
+
+def render_markdown(board: dict[str, Any]) -> str:
+    lines = [
+        "# Claw Code 2.0 Canonical Board",
+        "",
+        f"Generated: `{board['generated_at']}`",
+        f"Roadmap SHA-256 prefix: `{board['sources']['roadmap']['sha256_prefix']}`",
+        "",
+        "## Summary",
+        "",
+        f"- Total items: **{len(board['items'])}**",
+        f"- Roadmap headings covered: **{board['coverage']['roadmap_headings_total']} / {board['coverage']['roadmap_headings_mapped']}**",
+        f"- Roadmap ordered actions covered: **{board['coverage']['roadmap_actions_total']} / {board['coverage']['roadmap_actions_mapped']}**",
+        "",
+        "### By lifecycle status",
+        "",
+    ]
+    for status, count in board["summary"]["by_status"].items():
+        lines.append(f"- `{status}`: {count}")
+    lines.extend(["", "### By release bucket", ""])
+    for bucket, count in board["summary"]["by_release_bucket"].items():
+        lines.append(f"- `{bucket}`: {count}")
+    lines.extend(["", "## Board Items", ""])
+    for item in board["items"]:
+        deps = ", ".join(item.get("dependencies") or []) or "none"
+        rationale = item.get("deferral_rationale") or ""
+        lines.extend([
+            f"### {item['id']}",
+            f"- Title: {item['title']}",
+            f"- Source: `{item['source_anchor']}` (`{item['source_type']}`)",
+            f"- Bucket/status: `{item['release_bucket']}` / `{item['status']}`",
+            f"- Category/lane: `{item.get('category')}` / `{item.get('owner_lane')}`",
+            f"- Dependencies: {deps}",
+            f"- Verification: `{item['verification_required']}`",
+            f"- Deferral rationale: {rationale}",
+            "",
+        ])
+    return "\n".join(lines)
+
+
+def validate_board(board: dict[str, Any]) -> list[str]:
+    errors: list[str] = []
+    seen = set()
+    for index, item in enumerate(board.get("items", []), 1):
+        missing = [field for field in REQUIRED_ITEM_FIELDS if field not in item]
+        if missing:
+            errors.append(f"item {index} missing fields: {missing}")
+        if item.get("id") in seen:
+            errors.append(f"duplicate id: {item.get('id')}")
+        seen.add(item.get("id"))
+        if item.get("status") not in STATUSES:
+            errors.append(f"{item.get('id')} invalid status {item.get('status')}")
+        if item.get("release_bucket") not in RELEASE_BUCKETS:
+            errors.append(f"{item.get('id')} invalid release_bucket {item.get('release_bucket')}")
+        if not isinstance(item.get("dependencies"), list):
+            errors.append(f"{item.get('id')} dependencies must be list")
+    coverage = board.get("coverage", {})
+    if coverage.get("unmapped_roadmap_heading_lines"):
+        errors.append(f"unmapped heading lines: {coverage['unmapped_roadmap_heading_lines']}")
+    if coverage.get("duplicate_roadmap_heading_lines"):
+        errors.append(f"duplicate heading lines: {coverage['duplicate_roadmap_heading_lines']}")
+    if coverage.get("roadmap_headings_total") != coverage.get("roadmap_headings_mapped"):
+        errors.append("roadmap heading total/mapped mismatch")
+    return errors
+
+
+def build_board(repo_root: Path) -> dict[str, Any]:
+    roadmap_path = repo_root / "ROADMAP.md"
+    source_omx = find_source_omx(repo_root)
+    research = source_omx / "research"
+    plan_path = source_omx / "plans" / "claw-code-2-0-adaptive-plan.md"
+    headings, actions = parse_roadmap(roadmap_path)
+    items = [roadmap_item(record, i) for i, record in enumerate(headings, 1)]
+    items.extend(roadmap_item(record, i) for i, record in enumerate(actions, 1))
+
+    latest_issues = load_json(research / "claw-open-latest.json")
+    all_issues = load_json(research / "claw-issues.json")
+    items.extend(issue_item(issue, "claw-open-latest", "latest_open_issue", "2.x_intake") for issue in latest_issues)
+    # Include a small real-issue sample from the full freeze to keep the board tied to the larger issue manifest without exploding scope.
+    for issue in all_issues[:50]:
+        title_body = f"{issue.get('title','')} {issue.get('body','')}".lower()
+        if any(k in title_body for k in ["security", "windows", "install", "provider", "model", "session", "license", "zed", "spam", "plugin"]):
+            items.append(issue_item(issue, "claw-issues", "issue_theme", "beta_adoption"))
+    for source_name in ["opencode", "codex"]:
+        repo_meta = load_json(research / f"{source_name}-repo.json")
+        items.append(repo_context_item(repo_meta, source_name))
+
+    heading_lines = [record.line for record in headings]
+    mapped_heading_lines = [item["source_line"] for item in items if item.get("source_type") == "roadmap_heading"]
+    duplicate_heading_lines = sorted(line for line in set(mapped_heading_lines) if mapped_heading_lines.count(line) != 1)
+    unmapped_heading_lines = sorted(set(heading_lines) - set(mapped_heading_lines))
+
+    board = {
+        "schema_version": "cc2.board.v1",
+        "generated_at": datetime.now(timezone.utc).replace(microsecond=0).isoformat(),
+        "generation_policy": {
+            "ultragoal_mutation": "forbidden",
+            "roadmap_coverage": "all markdown headings plus top-level ordered roadmap actions",
+            "status_values": sorted(STATUSES),
+            "release_buckets": sorted(RELEASE_BUCKETS),
+        },
+        "sources": {
+            "roadmap": {
+                "path": "ROADMAP.md",
+                "sha256_prefix": sha256_prefix(roadmap_path),
+                "heading_count": len(headings),
+                "ordered_action_count": len(actions),
+            },
+            "approved_plan": {
+                "path": ".omx/plans/claw-code-2-0-adaptive-plan.md",
+                "sha256_prefix": sha256_prefix(plan_path),
+            },
+            "research": {
+                "root": str(source_omx / "research"),
+                "claw_open_latest_count": len(latest_issues),
+                "claw_issues_count": len(all_issues),
+                "opencode_repo": ".omx/research/opencode-repo.json",
+                "codex_repo": ".omx/research/codex-repo.json",
+            },
+        },
+        "coverage": {
+            "roadmap_headings_total": len(headings),
+            "roadmap_headings_mapped": len(mapped_heading_lines),
+            "unmapped_roadmap_heading_lines": unmapped_heading_lines,
+            "duplicate_roadmap_heading_lines": duplicate_heading_lines,
+            "roadmap_actions_total": len(actions),
+            "roadmap_actions_mapped": len([item for item in items if item.get("source_type") == "roadmap_action"]),
+        },
+        "summary": {},
+        "items": items,
+    }
+    board["summary"] = {
+        "by_status": summarize_counts(items, "status"),
+        "by_release_bucket": summarize_counts(items, "release_bucket"),
+        "by_source_type": summarize_counts(items, "source_type"),
+        "by_owner_lane": summarize_counts(items, "owner_lane"),
+    }
+    errors = validate_board(board)
+    if errors:
+        raise SystemExit("board validation failed:\n" + "\n".join(errors))
+    return board
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--repo-root", type=Path, default=Path.cwd())
+    parser.add_argument("--out-dir", type=Path, default=None)
+    args = parser.parse_args()
+    repo_root = args.repo_root.resolve()
+    out_dir = args.out_dir or (repo_root / ".omx" / "cc2")
+    out_dir.mkdir(parents=True, exist_ok=True)
+    board = build_board(repo_root)
+    board_json = out_dir / "board.json"
+    board_md = out_dir / "board.md"
+    board_json.write_text(json.dumps(board, indent=2, sort_keys=True) + "\n", encoding="utf-8")
+
+    renderer = repo_root / ".omx" / "cc2" / "render_board_md.py"
+    if renderer.exists():
+        subprocess.run([sys.executable, str(renderer), str(board_json), str(board_md)], check=True, cwd=str(repo_root))
+    else:
+        board_md.write_text(render_markdown(board) + "\n", encoding="utf-8")
+
+    print(f"wrote {board_json}")
+    print(f"wrote {board_md}")
+    print(f"roadmap headings mapped: {board['coverage']['roadmap_headings_mapped']}/{board['coverage']['roadmap_headings_total']}")
+    print(f"roadmap actions mapped: {board['coverage']['roadmap_actions_mapped']}/{board['coverage']['roadmap_actions_total']}")
+    return 0
+
+if __name__ == "__main__":
+    raise SystemExit(main())

scripts/validate_cc2_board.py

@@ -0,0 +1,87 @@
+#!/usr/bin/env python3
+"""Validate the generated Claw Code 2.0 board coverage and schema."""
+from __future__ import annotations
+
+import argparse
+import json
+import re
+from pathlib import Path
+
+REQUIRED = {
+    "id",
+    "title",
+    "source_anchor",
+    "source_type",
+    "release_bucket",
+    "status",
+    "dependencies",
+    "verification_required",
+    "deferral_rationale",
+}
+STATUSES = {
+    "context",
+    "active",
+    "open",
+    "done_verify",
+    "stale_done",
+    "superseded",
+    "deferred_with_rationale",
+    "rejected_not_claw",
+}
+
+def roadmap_heading_lines(path: Path) -> list[int]:
+    lines = []
+    for line_no, line in enumerate(path.read_text(encoding="utf-8").splitlines(), 1):
+        if re.match(r"^#{1,6}\s+", line):
+            lines.append(line_no)
+    return lines
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--repo-root", type=Path, default=Path.cwd())
+    parser.add_argument("--board", type=Path, default=None)
+    args = parser.parse_args()
+    repo_root = args.repo_root.resolve()
+    board_path = args.board or (repo_root / ".omx" / "cc2" / "board.json")
+    board = json.loads(board_path.read_text(encoding="utf-8"))
+    errors: list[str] = []
+    ids = set()
+    for index, item in enumerate(board.get("items", []), 1):
+        missing = REQUIRED - set(item)
+        if missing:
+            errors.append(f"item {index} missing required fields: {sorted(missing)}")
+        if item.get("id") in ids:
+            errors.append(f"duplicate id: {item.get('id')}")
+        ids.add(item.get("id"))
+        if item.get("status") not in STATUSES:
+            errors.append(f"{item.get('id')} invalid status {item.get('status')}")
+        if not isinstance(item.get("dependencies"), list):
+            errors.append(f"{item.get('id')} dependencies must be list")
+    expected = roadmap_heading_lines(repo_root / "ROADMAP.md")
+    mapped = [item.get("source_line") for item in board.get("items", []) if item.get("source_type") == "roadmap_heading"]
+    unmapped = sorted(set(expected) - set(mapped))
+    duplicates = sorted(line for line in set(mapped) if mapped.count(line) != 1)
+    if unmapped:
+        errors.append(f"unmapped ROADMAP headings: {unmapped}")
+    if duplicates:
+        errors.append(f"duplicate ROADMAP heading mappings: {duplicates}")
+    coverage = board.get("coverage", {})
+    if coverage.get("roadmap_headings_total") != len(expected):
+        errors.append("coverage roadmap_headings_total does not match ROADMAP.md")
+    if coverage.get("roadmap_headings_mapped") != len(mapped):
+        errors.append("coverage roadmap_headings_mapped does not match board items")
+    if errors:
+        print("FAIL cc2 board validation")
+        for error in errors:
+            print(f"- {error}")
+        return 1
+    print("PASS cc2 board validation")
+    print(f"- board: {board_path}")
+    print(f"- items: {len(board.get('items', []))}")
+    print(f"- ROADMAP headings mapped: {len(mapped)}/{len(expected)}")
+    print(f"- ROADMAP actions mapped: {coverage.get('roadmap_actions_mapped')}/{coverage.get('roadmap_actions_total')}")
+    return 0
+
+if __name__ == "__main__":
+    raise SystemExit(main())

src/models.py

@@ -23,6 +23,7 @@ class PortingModule:
 class PermissionDenial:
     tool_name: str
     reason: str
+    status: str = 'blocked'
 
 
 @dataclass(frozen=True)

src/path_scope.py

@@ -0,0 +1,150 @@
+from __future__ import annotations
+
+import glob
+import os
+import re
+import shlex
+from dataclasses import dataclass
+from pathlib import Path, PureWindowsPath
+
+_GLOB_META = set('*?[')
+_WINDOWS_DRIVE_RE = re.compile(r'^[A-Za-z]:[\\/]')
+_WINDOWS_UNC_RE = re.compile(r'^(?:\\\\|//)[^\\/]+[\\/][^\\/]+')
+_ENV_ASSIGNMENT_RE = re.compile(r'^[A-Za-z_][A-Za-z0-9_]*=')
+
+
+@dataclass(frozen=True)
+class PathScopeDecision:
+    allowed: bool
+    reason: str
+    candidate: str | None = None
+    resolved: str | None = None
+
+
+@dataclass(frozen=True)
+class WorkspacePathScope:
+    """Validate tool/shell path operands against explicit workspace roots.
+
+    The policy is intentionally conservative for the Python port: any candidate
+    path that resolves outside the configured roots is denied, including paths
+    reached through symlinks or glob expansion. Windows drive/UNC paths are
+    treated as out-of-scope on POSIX roots unless an allowed root is also a
+    Windows-style root with the same prefix.
+    """
+
+    roots: tuple[Path, ...]
+
+    @classmethod
+    def from_root(cls, root: str | Path) -> 'WorkspacePathScope':
+        return cls.from_roots((root,))
+
+    @classmethod
+    def from_roots(cls, roots: tuple[str | Path, ...] | list[str | Path]) -> 'WorkspacePathScope':
+        resolved_roots = tuple(Path(root).expanduser().resolve(strict=False) for root in roots)
+        if not resolved_roots:
+            raise ValueError('at least one workspace root is required')
+        return cls(resolved_roots)
+
+    def validate_payload(self, payload: str, cwd: str | Path | None = None) -> PathScopeDecision:
+        cwd_path = Path(cwd).expanduser().resolve(strict=False) if cwd else self.roots[0]
+        cwd_decision = self.validate_path(cwd_path)
+        if not cwd_decision.allowed:
+            return PathScopeDecision(False, f'cwd outside workspace scope: {cwd_path}', str(cwd_path), cwd_decision.resolved)
+        for candidate in extract_path_candidates(payload):
+            decision = self.validate_path(candidate, cwd_path)
+            if not decision.allowed:
+                return decision
+        return PathScopeDecision(True, 'all path candidates are inside workspace scope')
+
+    def validate_path(self, candidate: str | Path, cwd: str | Path | None = None) -> PathScopeDecision:
+        raw = os.path.expandvars(os.path.expanduser(str(candidate)))
+        if _is_windows_absolute(raw):
+            return self._validate_windows_path(raw)
+        base = Path(cwd).expanduser().resolve(strict=False) if cwd else self.roots[0]
+        path = Path(raw)
+        if not path.is_absolute():
+            path = base / path
+        expanded = self._expand_glob(path)
+        for expanded_path in expanded:
+            resolved = expanded_path.resolve(strict=False)
+            if not any(_is_relative_to(resolved, root) for root in self.roots):
+                return PathScopeDecision(
+                    False,
+                    'path resolves outside workspace scope',
+                    str(candidate),
+                    str(resolved),
+                )
+        return PathScopeDecision(True, 'path is inside workspace scope', str(candidate), str(expanded[0].resolve(strict=False)))
+
+    def _expand_glob(self, path: Path) -> tuple[Path, ...]:
+        path_text = str(path)
+        if any(char in path_text for char in _GLOB_META):
+            matches = tuple(Path(match) for match in glob.glob(path_text, recursive=True))
+            if matches:
+                return matches
+            # For unmatched globs, validate the stable non-glob parent prefix.
+            stable_parts: list[str] = []
+            for part in path.parts:
+                if any(char in part for char in _GLOB_META):
+                    break
+                stable_parts.append(part)
+            if stable_parts:
+                return (Path(*stable_parts),)
+        return (path,)
+
+    def _validate_windows_path(self, raw: str) -> PathScopeDecision:
+        candidate = PureWindowsPath(raw)
+        for root in self.roots:
+            root_text = str(root)
+            if not _is_windows_absolute(root_text):
+                continue
+            try:
+                candidate.relative_to(PureWindowsPath(root_text))
+                return PathScopeDecision(True, 'windows path is inside workspace scope', raw, str(candidate))
+            except ValueError:
+                continue
+        return PathScopeDecision(False, 'windows absolute path is outside workspace scope', raw, str(candidate))
+
+
+def extract_path_candidates(payload: str) -> tuple[str, ...]:
+    """Return conservative path-like operands from a shell/tool payload."""
+
+    try:
+        tokens = shlex.split(payload, posix=True)
+    except ValueError:
+        tokens = payload.split()
+    raw_tokens = payload.split()
+    candidates: list[str] = []
+    for token in (*tokens, *raw_tokens):
+        if not token or token.startswith('-') or _ENV_ASSIGNMENT_RE.match(token):
+            continue
+        expanded = os.path.expandvars(os.path.expanduser(token))
+        if _looks_like_path(token) or _looks_like_path(expanded):
+            candidate = expanded if _looks_like_path(expanded) else token
+            if candidate not in candidates:
+                candidates.append(candidate)
+    return tuple(candidates)
+
+
+def _looks_like_path(token: str) -> bool:
+    return (
+        token in {'.', '..'}
+        or token.startswith(('./', '../', '/', '~/', '~/'))
+        or '..' in token.split('/')
+        or '/' in token
+        or '\\' in token
+        or any(char in token for char in _GLOB_META)
+        or _is_windows_absolute(token)
+    )
+
+
+def _is_windows_absolute(value: str) -> bool:
+    return bool(_WINDOWS_DRIVE_RE.match(value) or _WINDOWS_UNC_RE.match(value))
+
+
+def _is_relative_to(path: Path, root: Path) -> bool:
+    try:
+        path.relative_to(root)
+        return True
+    except ValueError:
+        return False

src/permissions.py

@@ -1,20 +1,49 @@
 from __future__ import annotations
 
 from dataclasses import dataclass, field
+from pathlib import Path
+
+from .path_scope import PathScopeDecision, WorkspacePathScope
 
 
 @dataclass(frozen=True)
 class ToolPermissionContext:
     deny_names: frozenset[str] = field(default_factory=frozenset)
     deny_prefixes: tuple[str, ...] = ()
+    workspace_scope: WorkspacePathScope | None = None
+    cwd: Path | None = None
 
     @classmethod
-    def from_iterables(cls, deny_names: list[str] | None = None, deny_prefixes: list[str] | None = None) -> 'ToolPermissionContext':
+    def from_iterables(
+        cls,
+        deny_names: list[str] | None = None,
+        deny_prefixes: list[str] | None = None,
+        workspace_root: str | Path | None = None,
+        workspace_roots: list[str | Path] | tuple[str | Path, ...] | None = None,
+        cwd: str | Path | None = None,
+    ) -> 'ToolPermissionContext':
+        roots: list[str | Path] = []
+        if workspace_roots:
+            roots.extend(workspace_roots)
+        if workspace_root is not None:
+            roots.append(workspace_root)
         return cls(
             deny_names=frozenset(name.lower() for name in (deny_names or [])),
             deny_prefixes=tuple(prefix.lower() for prefix in (deny_prefixes or [])),
+            workspace_scope=WorkspacePathScope.from_roots(roots) if roots else None,
+            cwd=Path(cwd).expanduser().resolve(strict=False) if cwd is not None else None,
         )
 
     def blocks(self, tool_name: str) -> bool:
         lowered = tool_name.lower()
         return lowered in self.deny_names or any(lowered.startswith(prefix) for prefix in self.deny_prefixes)
+
+    def validate_payload_scope(self, tool_name: str, payload: str) -> PathScopeDecision:
+        if self.workspace_scope is None or not _scope_checked_tool(tool_name):
+            return PathScopeDecision(True, 'workspace path scope not required for this tool')
+        return self.workspace_scope.validate_payload(payload, cwd=self.cwd)
+
+
+def _scope_checked_tool(tool_name: str) -> bool:
+    lowered = tool_name.lower()
+    return any(marker in lowered for marker in ('bash', 'shell', 'powershell', 'fileread', 'filewrite', 'fileedit'))

src/query_engine.py

@@ -82,6 +82,7 @@ class QueryEnginePort:
             f'Matched commands: {", ".join(matched_commands) if matched_commands else "none"}',
             f'Matched tools: {", ".join(matched_tools) if matched_tools else "none"}',
             f'Permission denials: {len(denied_tools)}',
+            *(f'Permission denial: {denial.tool_name} status={denial.status} reason={denial.reason}' for denial in denied_tools),
         ]
         output = self._format_output(summary_lines)
         projected_usage = self.total_usage.add_turn(prompt, output)
@@ -116,7 +117,13 @@ class QueryEnginePort:
         if matched_tools:
             yield {'type': 'tool_match', 'tools': matched_tools}
         if denied_tools:
-            yield {'type': 'permission_denial', 'denials': [denial.tool_name for denial in denied_tools]}
+            yield {
+                'type': 'permission_denial',
+                'denials': [
+                    {'tool_name': denial.tool_name, 'reason': denial.reason, 'status': denial.status}
+                    for denial in denied_tools
+                ],
+            }
         result = self.submit_message(prompt, matched_commands, matched_tools, denied_tools)
         yield {'type': 'message_delta', 'text': result.output}
         yield {

src/tools.py

@@ -78,10 +78,25 @@ def find_tools(query: str, limit: int = 20) -> list[PortingModule]:
     return matches[:limit]
 
 
-def execute_tool(name: str, payload: str = '') -> ToolExecution:
+def execute_tool(name: str, payload: str = '', permission_context: ToolPermissionContext | None = None) -> ToolExecution:
     module = get_tool(name)
     if module is None:
         return ToolExecution(name=name, source_hint='', payload=payload, handled=False, message=f'Unknown mirrored tool: {name}')
+    if permission_context and permission_context.blocks(module.name):
+        return ToolExecution(name=module.name, source_hint=module.source_hint, payload=payload, handled=False, message=f"Permission denied for mirrored tool '{module.name}'.")
+    if permission_context:
+        scope_decision = permission_context.validate_payload_scope(module.name, payload)
+        if not scope_decision.allowed:
+            return ToolExecution(
+                name=module.name,
+                source_hint=module.source_hint,
+                payload=payload,
+                handled=False,
+                message=(
+                    f"Permission denied for mirrored tool '{module.name}': {scope_decision.reason}"
+                    f" (candidate={scope_decision.candidate!r}, resolved={scope_decision.resolved!r})."
+                ),
+            )
     action = f"Mirrored tool '{module.name}' from {module.source_hint} would handle payload {payload!r}."
     return ToolExecution(name=module.name, source_hint=module.source_hint, payload=payload, handled=True, message=action)
 

tests/test_security_scope.py

@@ -0,0 +1,135 @@
+from __future__ import annotations
+
+import os
+import tempfile
+import unittest
+from pathlib import Path
+
+from src.models import PermissionDenial
+from src.path_scope import WorkspacePathScope, extract_path_candidates
+from src.permissions import ToolPermissionContext
+from src.query_engine import QueryEnginePort
+from src.tools import execute_tool
+
+
+class WorkspacePathScopeTests(unittest.TestCase):
+    def test_direct_parent_escape_is_denied(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            workspace = Path(tmp) / 'workspace'
+            workspace.mkdir()
+            decision = WorkspacePathScope.from_root(workspace).validate_payload('cat ../secret.txt')
+            self.assertFalse(decision.allowed)
+            self.assertIn('outside workspace scope', decision.reason)
+
+    def test_issue_3007_symlink_escape_is_denied(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            workspace = root / 'workspace'
+            outside = root / 'outside'
+            workspace.mkdir()
+            outside.mkdir()
+            (outside / 'secret.txt').write_text('secret')
+            link = workspace / 'linked-outside'
+            link.symlink_to(outside, target_is_directory=True)
+
+            decision = WorkspacePathScope.from_root(workspace).validate_payload('cat linked-outside/secret.txt')
+
+            self.assertFalse(decision.allowed)
+            self.assertIn(str(outside.resolve()), decision.resolved or '')
+
+    def test_glob_expansion_must_stay_inside_workspace(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            workspace = root / 'workspace'
+            outside = root / 'outside'
+            workspace.mkdir()
+            outside.mkdir()
+            (outside / 'secret.txt').write_text('secret')
+
+            decision = WorkspacePathScope.from_root(workspace).validate_payload(f'cat {outside}/*.txt')
+
+            self.assertFalse(decision.allowed)
+            self.assertEqual(str((outside / 'secret.txt').resolve()), decision.resolved)
+
+    def test_shell_environment_expansion_is_validated(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            workspace = root / 'workspace'
+            outside = root / 'outside'
+            workspace.mkdir()
+            outside.mkdir()
+            previous = os.environ.get('CLAW_SCOPE_OUTSIDE')
+            os.environ['CLAW_SCOPE_OUTSIDE'] = str(outside)
+            try:
+                self.assertEqual((f'{outside}/secret.txt',), extract_path_candidates('cat $CLAW_SCOPE_OUTSIDE/secret.txt'))
+                decision = WorkspacePathScope.from_root(workspace).validate_payload('cat $CLAW_SCOPE_OUTSIDE/secret.txt')
+            finally:
+                if previous is None:
+                    os.environ.pop('CLAW_SCOPE_OUTSIDE', None)
+                else:
+                    os.environ['CLAW_SCOPE_OUTSIDE'] = previous
+
+            self.assertFalse(decision.allowed)
+            self.assertIn(str(outside.resolve()), decision.resolved or '')
+
+    def test_explicit_worktree_roots_are_allowed(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            workspace = root / 'workspace'
+            worktree = root / 'worktree'
+            workspace.mkdir()
+            worktree.mkdir()
+            (worktree / 'file.txt').write_text('ok')
+
+            decision = WorkspacePathScope.from_roots((workspace, worktree)).validate_payload(f'cat {worktree}/file.txt')
+
+            self.assertTrue(decision.allowed, decision.reason)
+
+    def test_windows_absolute_paths_are_denied_for_posix_workspace(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            workspace = Path(tmp) / 'workspace'
+            workspace.mkdir()
+
+            drive_decision = WorkspacePathScope.from_root(workspace).validate_payload(r'type C:\Users\other\secret.txt')
+            unc_decision = WorkspacePathScope.from_root(workspace).validate_payload(r'type \\server\share\secret.txt')
+
+            self.assertFalse(drive_decision.allowed)
+            self.assertIn('windows absolute path', drive_decision.reason)
+            self.assertFalse(unc_decision.allowed)
+            self.assertIn('windows absolute path', unc_decision.reason)
+
+    def test_file_and_shell_tools_use_workspace_scope_context(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            workspace = root / 'workspace'
+            outside = root / 'outside'
+            workspace.mkdir()
+            outside.mkdir()
+            context = ToolPermissionContext.from_iterables(workspace_root=workspace, cwd=workspace)
+
+            file_result = execute_tool('FileReadTool', f'{outside}/secret.txt', permission_context=context)
+            shell_result = execute_tool('BashTool', f'cat {outside}/secret.txt', permission_context=context)
+            inside_result = execute_tool('FileReadTool', './allowed.txt', permission_context=context)
+
+            self.assertFalse(file_result.handled)
+            self.assertIn('Permission denied', file_result.message)
+            self.assertFalse(shell_result.handled)
+            self.assertIn('Permission denied', shell_result.message)
+            self.assertTrue(inside_result.handled)
+
+    def test_permission_denial_stream_events_expose_status_and_reason(self) -> None:
+        engine = QueryEnginePort.from_workspace()
+        denial = PermissionDenial('BashTool', 'path resolves outside workspace scope')
+
+        events = list(engine.stream_submit_message('cat ../secret.txt', matched_tools=('BashTool',), denied_tools=(denial,)))
+        permission_event = next(event for event in events if event['type'] == 'permission_denial')
+        result = engine.submit_message('cat ../secret.txt', matched_tools=('BashTool',), denied_tools=(denial,))
+
+        self.assertEqual('blocked', permission_event['denials'][0]['status'])
+        self.assertEqual('path resolves outside workspace scope', permission_event['denials'][0]['reason'])
+        self.assertIn('status=blocked', result.output)
+        self.assertIn('path resolves outside workspace scope', result.output)
+
+
+if __name__ == '__main__':
+    unittest.main()