ship-loop-test

name	ship-loop-test
description	Jest test suite child loop. Runs `npm test` (uses jest.worktree.cjs per package.json), records pass/fail counts and failing suite paths into ship-loop-state.json. Only fires if last ship-loop-build was green. Applies one recovery recipe (jest config / mock generation / worktree cleanup) on failure, retries once, escalates. Use standalone via /loop ship-loop-test or wired into the master orchestrator.
type	child-loop
context	persistent

ship-loop-test — jest test suite child loop

Activation

• Standalone: /loop ship-loop-test
• Orchestrated: invoked by ship-loop-master when build.state === 'green' AND (test.state !== 'green' OR test.last_run > 30 min ago)

Pre-condition gate

Read ship-loop-state.json. If layers.build.state !== 'green':

• Skip this run; write event {"reason":"build_not_green","skipped":true}
• Return to master without changing test state

Process

Step 1: Run jest

cd D:/Synthex
npm test 2>&1 | tee /tmp/ship-loop-test-output.log

Parse output for:

• Test Suites: X failed, Y skipped, Z passed, T total line
• Tests: A failed, B skipped, C todo, D passed, E total line
• All ^FAIL lines (failing test file paths)
• For each failure, extract first 5 lines of error context

Step 2: Update state

Atomic update to layers.test:

{
  "state": "green" | "red",
  "last_run": "<iso>",
  "retries": <int>,
  "details": {
    "suites": { "passed": <int>, "failed": <int>, "skipped": <int>, "total": <int> },
    "tests": { "passed": <int>, "failed": <int>, "todo": <int>, "skipped": <int>, "total": <int> },
    "failing_suites": [
      { "path": "tests/...", "first_error": "..." }
    ]
  }
}

state === 'green' IFF suites.failed === 0 AND tests.failed === 0.

Step 3: Recovery sub-loop on red

If state === 'red' AND retries < 1:

For each failing suite (process up to 3 per tick to avoid runaway):

1. Match error against recovery-recipes.md (priorities below)
2. If match: apply recipe, increment retries, re-run JUST that suite via npx jest --config jest.worktree.cjs <path>, update state
3. If no match OR retry-also-fails: append to escalations.md:

   ## P1 — Test failure: <suite path>
   
   - When: <iso>
   - First error: <verbatim>
   - Recipe attempted: <# or "no match">
   - Repro: `cd D:/Synthex && npx jest --config jest.worktree.cjs <path>`
   

Recipe priorities for this loop

Most likely matches:

• #3 (uncrypto/ESM) — SyntaxError: Unexpected token 'export'
• #6 (fail() deprecation) — ReferenceError: fail is not defined
• #7 (auth ratchet breach) — Auth coverage ratchet breached
• #9 (worktree pollution) — failing path starts with .claude/worktrees/

Bespoke recovery: missing mock

If error matches Cannot find module '<pkg>' AND <pkg> is in package.json AND a similar mock exists at tests/__mocks__/:

• Auto-create tests/__mocks__/<pkg>.js with skeleton:

  // Auto-generated mock by ship-loop-test recovery
  module.exports = {};
  module.exports.default = {};
  

• Re-run; if still fails, escalate (mock needs hand-curation)

Verification

• Insert a failing test (expect(true).toBe(false)) in tests/unit/foo.test.ts; expect first run red, no recipe match, escalation written
• Add import 'uncrypto' to a test file and remove uncrypto from transformIgnorePatterns; expect recipe #3 to match, recipe applied, retry succeeds

Out of scope

• Coverage threshold validation (defer to CI; jest config already has thresholds)
• Flaky test detection (would need re-run statistics — Phase 2 enhancement)
• E2E playwright tests (excluded per testPathIgnorePatterns in jest configs)