Spec Kitty

└─ kitty-specs
   └─ Workflow Parity Fixes 988/989/991

Mission Run:

📚 Docs ↗

Data Model — Workflow Parity Fixes 988/989/991

New Types

`ClaimablePreview` (in `src/specify_cli/next/`)

A frozen dataclass returned by a side-effect-free claim-discovery helper used by next --json.

Field	Type	Description	Invariant
`wp_id`	`str \	None`	The WP that `agent action implement` would claim, or `None` when no candidate can be selected.
`selection_reason`	`str \	None`	A short stable token identifying why selection was suppressed. Examples: `"no_planned_wps"`, `"all_wps_in_progress"`, `"dependencies_unsatisfied"`, `"baseline_violation"`.
`candidates`	`tuple[str, ...]`	Ordered list of WP IDs the claim algorithm would have considered.	Deterministic by lane/order; identical to the order `start_implementation_status()` would have walked.

Lifetime: per-invocation only; never persisted.

New Diagnostic Codes

`LIGHTWEIGHT_REVIEW_MISSING_BASELINE` (in `src/specify_cli/cli/commands/review/_diagnostics.py`)

Emitted by scan_dead_code() when called from the lightweight review path for a modern mission (one whose meta.json has a populated mission_id) whose baseline_merge_commit is null.

Payload field	Type	Description
`code`	`str`	Always `"LIGHTWEIGHT_REVIEW_MISSING_BASELINE"`.
`mission_id`	`str`	The mission's canonical ULID.
`mission_slug`	`str`	The mission's human slug.
`remediation`	`str`	Stable hint string; for tests, the substring `"baseline_merge_commit"` MUST appear.

`LEGACY_MISSION_DEAD_CODE_SKIP` (optional, in `_diagnostics.py`)

Emitted in the legacy path so that the silent-skip behavior remains greppable. The lightweight review still returns a passing verdict on legacy missions but tags the verdict with this code so it cannot be confused with a true clean pass.

Reused Types (no schema change)

MissionMeta — read-only consumers are the only change point; the mission_id field already exists.
MissionReviewMode (src/specify_cli/cli/commands/review/_mode.py) — unchanged.
WorkPackageClaimConflict (src/specify_cli/status/work_package_lifecycle.py) — unchanged; the new helper does not raise this because it does not attempt to claim.
find_rejected_review_artifact_conflicts(...) return shape — unchanged; just called from a new caller (dry-run).
REJECTED_REVIEW_ARTIFACT_CONFLICT constant — unchanged.

State Transitions

No new lane transitions are introduced. The 9-lane state machine (planned → claimed → in_progress → for_review → in_review → approved → done, plus blocked/canceled) is unchanged. The claim-discovery helper performs no transition — it computes the same candidate set start_implementation_status() would have used, then returns it.

Invariants

I-001: ClaimablePreview.wp_id is None XOR selection_reason is None — exactly one is set.
I-002: LIGHTWEIGHT_REVIEW_MISSING_BASELINE is emitted only when mission_id is present in meta.json and baseline_merge_commit is null.
I-003: merge --dry-run invokes find_rejected_review_artifact_conflicts() exactly once per dry-run invocation, before any merge preview output is emitted.