docs(roadmap): add #424 — init json emits dual artifact schema: artifacts[] and flat created/skipped/updated

ROADMAP.md

@@ -6303,4 +6303,4 @@ Original filing (2026-04-18): the session emitted `SessionStart hook (completed)
 
 422. **`export --output-format json` and `--resume latest` report the same "no managed sessions" scenario using two different `kind` codes — `no_managed_sessions` vs `session_load_failed` — making "no session found" undetectable by a single kind-code check** — dogfooded 2026-04-30 KST (UTC+9) by Jobdori on `e939777f`. Running `claw export --output-format json` with no session present returns (on stderr, exit 1): `{"error":"no managed sessions found in .claw/sessions/<fingerprint>/","hint":"Start \`claw\` to create a session, then rerun with \`--resume latest\`.\nNote: claw partitions sessions per workspace fingerprint; sessions from other CWDs are invisible.","kind":"no_managed_sessions","type":"error"}`. Running `claw --resume latest /status --output-format json` with no session present returns (on stderr, exit 1): `{"error":"failed to restore session: no managed sessions found in .claw/sessions/<fingerprint>/","hint":"Start \`claw\` to create a session, then rerun with \`--resume latest\`.\nNote: claw partitions sessions per workspace fingerprint; sessions from other CWDs are invisible.","kind":"session_load_failed","type":"error"}`. Both describe the same root condition — there are no sessions to operate on — but they expose it via different `kind` discriminants. Automation that checks `kind == "no_managed_sessions"` to detect a cold workspace will miss the `--resume` path's `session_load_failed`, and vice versa. A wrapper that guards "run with --resume only if a session exists" must special-case both codes. The hint text is identical between them, suggesting the messages are logically equivalent. Additionally neither code matches the proposed canonical names `session_not_found` / `session_load_failed` as stable `ErrorKind` discriminants described in ROADMAP #77's fix shape, which explicitly proposes typed error-kind codes for session lifecycle failures. **Required fix shape:** (a) unify "no sessions found for this workspace fingerprint" under a single canonical `kind` code — either `no_managed_sessions` or `session_not_found` — used consistently by every command path that encounters an empty session registry; (b) if `session_load_failed` is a more general category (covering e.g. corrupt session files, IO errors, schema version mismatches), it should nest a concrete `reason:"no_managed_sessions"` or `reason:"session_not_found"` sub-field so callers can distinguish "empty registry" from "found but unreadable"; (c) align with the canonical error-kind contract proposed in #77; (d) add regression coverage proving `export` and `--resume latest` in an empty workspace both return an error with the same top-level `kind` code. **Why this matters:** session guard-rails in orchestration need a single stable `kind` to detect cold workspaces without enumerating all possible no-session synonyms. Two divergent codes for the same condition make defensive automation brittle and contradict the promise of machine-readable error envelopes. Source: Jobdori live dogfood, `e939777f`, 2026-04-30 KST (UTC+9).
 
-426. **`claw plugins --output-format json` surface has zero test coverage in `output_format_contract.rs` — the contract tests that lock every other major JSON surface (`agents`, `mcp`, `skills`, `status`, `sandbox`, `doctor`, `help`, `version`, `acp`, `bootstrap-plan`, `system-prompt`, `init`, `diff`, `config`) do not include a single assertion about `plugins`, leaving the plugin JSON shape entirely untested and free to silently drift** — dogfooded 2026-05-01 KST (UTC+9) by Jobdori on `57096b0a`. Inspection of `rust/crates/rusty-claude-cli/tests/output_format_contract.rs` (11 tests) shows coverage for every major `--output-format json` surface except `plugins`. Live binary produces `{"action":"list","kind":"plugin","message":"<prose table>","reload_runtime":false,"target":null}` for `claw plugins --output-format json`, but no test asserts (a) that `kind == "plugin"`, (b) that `action == "list"`, (c) that the payload is valid JSON at all, (d) that `reload_runtime` is present and boolean, or (e) that `target` is present and null/typed. The only plugin-related test in `mock_parity_harness.rs` covers plugin tool execution round-trip, not the CLI inspection surface. This means the `plugins list` JSON schema documented in ROADMAP #416 (mutation-shape envelope, missing `plugins:[]` array) can silently regress or change shape between commits with no CI signal. By contrast, if `agents list` drops the `agents[]` field, `inventory_commands_emit_structured_json_when_requested` catches it immediately. **Required fix shape:** (a) add a `plugins_command_emits_json_when_requested` test in `output_format_contract.rs` that: runs `claw plugins --output-format json` in an isolated temp dir, asserts stdout is valid JSON, asserts `kind == "plugin"`, asserts `action == "list"`, asserts `reload_runtime` is boolean, asserts `target` is JSON null when no plugin is targeted; (b) add a second assertion variant for `claw plugins enable <name> --output-format json` to confirm the lifecycle-mutation envelope fields are stable; (c) pair with the existing ROADMAP #416 fix so when `plugins:[]` array is added, a test assertion locks it; (d) extend to `plugins disable`, `plugins install`, and `plugins uninstall` shapes once they are implemented. **Why this matters:** test brittleness — the plugin JSON surface is the only major `--output-format json` surface with zero contract coverage. Any schema change, regression, or panic in `plugins` dispatch goes undetected until a downstream claw hits it at runtime. CI's four-check suite gives a false assurance of coverage for plugin JSON because `cargo test --workspace` passes even with zero plugin contract assertions. Source: Jobdori dogfood + test-file inspection, `57096b0a`, 2026-05-01 KST (UTC+9). Cross-reference: #416 (plugins list returns mutation shape, no plugins array).
+424. **`init --output-format json` emits the same artifact-state data in two parallel schemas simultaneously — `artifacts:[{name, status}]` and flat `created:[...]` / `skipped:[...]` / `updated:[...]` arrays — with no documented relationship, no deprecation signal, and no way to tell which is canonical** — dogfooded 2026-04-30 KST (UTC+9) by Jobdori on `e939777f`. Running `claw init --output-format json` in a partially-initialized project returns: `{"kind":"init","project_path":"/tmp/probe-clone","created":[".claw/"],"skipped":[".claw.json",".gitignore","CLAUDE.md"],"updated":[],"artifacts":[{"name":".claw/","status":"created"},{"name":".claw.json","status":"skipped"},{"name":".gitignore","status":"skipped"},{"name":"CLAUDE.md","status":"skipped"}],"next_step":"Review and tailor the generated guidance","message":"Init\n  Project  /tmp/probe-clone\n  .claw/  created\n  ..."}`. The `artifacts` array and the three flat arrays (`created`, `skipped`, `updated`) encode the same information: `.claw/` appears in both `created[0]` and `artifacts[0].status=="created"`. A claw consuming this envelope cannot tell which schema is the stable contract: `artifacts[].status` supports the full three-way distinction (created/skipped/updated) in one traversal, while `created`/`skipped`/`updated` require three separate lookups and cannot represent a file whose status changes atomically. Neither schema includes `schema_version`, deprecation metadata, or a "prefer this field" signal. Automation that starts with `artifacts[]` and later drops it in favor of the flat arrays (or vice versa) will silently double-count or miss artifacts on a version boundary. This is the post-fix state of ROADMAP #79: #79 documented that init shipped only prose `message`; the fix that added structured fields added both schemas simultaneously without reconciling them. **Required fix shape:** (a) designate one schema as canonical — `artifacts:[{name, status}]` is the richer representation; (b) deprecate `created`/`skipped`/`updated` flat arrays with a `deprecated:true` or `schema_version` signal, or remove them if no downstream consumer has stabilized on them; (c) if both must coexist for backward compatibility, document their equivalence and add a note that both will be kept in sync; (d) add regression coverage proving exactly one schema is marked canonical and that any deprecated fields carry an explicit deprecation signal. **Why this matters:** dual parallel schemas for the same init artifact state create ambiguity for orchestrators: both schemas must be parsed defensively, and any future field addition to one schema must be mirrored to the other or the divergence silently grows. Source: Jobdori live dogfood, `e939777f`, 2026-04-30 KST (UTC+9). Related: ROADMAP #79 (original prose-only init JSON, now fixed but left with dual schema residue).