T-LIYW-test-fixture-id-drift

Status: closed/done · Impact: medium · Complexity: small

Goal

The task schema (schema_version 5, plugin/lib/model/entities/task/schema.json) now lists id (pattern ^T-[0-9A-Z]{4}$) as a required property. Several test files still build task frontmatter — inline strings and fixture files — from before id was required, so those fixtures fail schema validation with 'id' is a required property. This is fixture drift, not a behavior bug: the production validator, skills, and schema are correct; only the test inputs are stale. Repairing them restores a clean test baseline so the milestone’s test gates run green.

Today

Location	Role today
`plugin/validators/tests/task_autonomy_enum.test.ts`	Builds task frontmatter without `id`; autonomy-enum cases fail schema validation.
`plugin/skills/task-auto-define/tests/task_auto_define.test.ts`	Seeds a task fixture lacking `id`; the auto-define readiness case fails.
`plugin/skills/task-define/tests/task_define.test.ts`	Inline frontmatter fixtures omit `id`; behavioural-layer cases report a spurious `<root>` failing pointer.
`plugin/skills/task-ensure-ready/tests/ensure_ready.test.ts`	Several fixtures omit `id`; readiness cases fail on the schema gate.
`plugin/skills/task-close-out/tests/prs_verify.test.ts`	Task fixtures used by the prs verify/append cases lack `id`.

Proposed

Every test fixture / inline task-frontmatter input across those five files carries a schema-valid, obviously-synthetic id (e.g. T-0000, T-TST1 matching ^T-[0-9A-Z]{4}$, plus any other field the schema now requires but the fixture omits). All five files run fully green. No production code, schema, skill .ts logic, or SKILL.md is touched. The full-suite failure count drops by exactly the fixture-drift amount these five files contribute.

Approach

Reproduce: bun test to capture the baseline, then per-file to isolate each failing set and confirm the error is 'id' is a required property (or other newly-required-but-missing schema fields), not behavior drift.
For each file, add a synthetic schema-valid id to every task fixture / inline frontmatter block that lacks one, keeping ids obviously fake.
Re-run each file to green, then the full suite to confirm the count drop.
Keep bunx tsc --noEmit green.

Files to touch

Location	Kind	Change
`plugin/validators/tests/task_autonomy_enum.test.ts`	modify	Add synthetic `id` to task frontmatter fixtures.
`plugin/skills/task-auto-define/tests/task_auto_define.test.ts`	modify	Add synthetic `id` to the seeded task fixture.
`plugin/skills/task-define/tests/task_define.test.ts`	modify	Add synthetic `id` to inline frontmatter fixtures.
`plugin/skills/task-ensure-ready/tests/ensure_ready.test.ts`	modify	Add synthetic `id` to the fixtures missing it.
`plugin/skills/task-close-out/tests/prs_verify.test.ts`	modify	Add synthetic `id` to the task fixtures.

Acceptance criteria

AC-1: bun test plugin/validators/tests/task_autonomy_enum.test.ts passes.
AC-2: The four skill test files (task-auto-define, task-define, task-ensure-ready/ensure_ready, task-close-out/prs_verify) pass.
AC-3: The full-suite failure count drops by exactly the fixture-drift amount these five files contribute, with no new failures introduced.
AC-4: bunx tsc --noEmit stays green.

Out of scope

Production code, the task schema, skill .ts logic, and SKILL.md files.
The tests/parity/validators.golden.test.ts golden-suite failures, which are driven by live docs/planning/ entity renames/migration owned by another agent — not by the five fixture sets named here.

Dependencies

none

Discovery context

Surfaced while preparing a clean test baseline for the active milestone: on clean main bun test reports a block of 'id' is a required property failures concentrated in these five test files after the task schema made id required at schema_version 5.

← Back to Tasks