Systematic Test Matrix

Run POSCursor verification surfaces in a fixed order. Keep evidence in tmp/systematic-test-matrix/<timestamp>/. Never claim green from stale logs.

When to use

Need	Action
Release / deploy confidence	This skill
Author or fix one Jest file	`testing` skill
Deploy to Cloudflare after green	`deploy-poscursor`
Fast iteration	`npx jest <path> --no-coverage` only

Quick start

Confirm cwd is the POSCursor repo root (posCURSORv2 / this workspace).
Run:

node .agents/skills/systematic-test-matrix/scripts/run_matrix.mjs

Open the newest tmp/systematic-test-matrix/<timestamp>/summary.json and any failing *.log before reporting results.

Workflow (script order)

1. Baseline (bail on failure by default)

npx tsc --noEmit
npx expo-doctor

2. Jest by discovered `tests` buckets

The script walks app/ and src/, finds __tests__ dirs with *.test.* files, and runs each with:

npx jest --runInBand --forceExit --runTestsByPath <files…>

Buckets are written to jest-buckets.json in the report dir. Typical buckets include services, utils, components, hooks, models, store(s), cic, and role screens under app/(admin), app/(cashier), app/(borrower), etc.

3. Integrated surfaces

jest_full — npm test -- --runInBand --forceExit (catches tests outside __tests__/ folders, e.g. *.regression.test.ts)
playwright_full — npx playwright test --workers=1 --reporter=list (skipped with --skip-playwright)
export_web — npm run export:web (script uses this, not export:web:production)

4. Export smoke (default on)

Serves dist/ on localhost, opens Chromium, fails on page/console errors. Writes smoke.json. Skip with --skip-smoke.

Commands

# Full matrix
node .agents/skills/systematic-test-matrix/scripts/run_matrix.mjs

# Debug without long E2E / smoke
node .agents/skills/systematic-test-matrix/scripts/run_matrix.mjs --skip-playwright --skip-smoke

# Resume one step
node .agents/skills/systematic-test-matrix/scripts/run_matrix.mjs --only=playwright_full
node .agents/skills/systematic-test-matrix/scripts/run_matrix.mjs --only=jest_full,export_web

# Collect all failures (no early exit)
node .agents/skills/systematic-test-matrix/scripts/run_matrix.mjs --no-bail

Outputs

Artifact	Meaning
`summary.json`	Per-step ok/exitCode/logPath; overall `ok`
`jest-buckets.json`	Discovered buckets for that run
`*.log`	Full command output
`smoke.json`	Export smoke URL/title/errors

Agent reporting rules

Quote the report directory path used.
For failures: step name, exit code, log path, first actionable error.
Do not mark release-ready if summary.ok is false.
Optional extra gates (lint, production export, sync integrity) are not in the default script — see references/test-surfaces.md and deploy-poscursor when shipping.

Unit/integration authoring: ../testing/SKILL.md
Deploy gates: ../deploy-poscursor/SKILL.md
Surface catalog: references/test-surfaces.md