agentleFS
genex-gamesgenex-games/genex-desktop/.agents/skills/run-area-tests/SKILL.md

run-area-tests

Skill874 starsChanged 6 days agoMITView on GitHub

Pick and run only the tests a change can affect - the affected-tests selector, one knowledge-map area, or named files - instead of the full suite. Use after each red/green step and before finishing a task step.

The file

Files
SKILL.md

Run the affected and area tests

Choose the cheapest layer that proves the behaviour; see docs/agent/verification.md for the full L0-L5 table. Never describe a focused pass as a full-suite pass.

The scope table owns required checks. Its layer catalog describes available commands, not a cumulative checklist. For bounded core/harness/engine changes, run the affected L3 suites; harness changes also need the incident gate. Reserve full verification for broad integration changes or explicit requests.

Selector

scripts/affected-tests.mjs is read-only. It takes changed files from git diff against the merge-base with origin/dev plus untracked files, walks each test's import graph, and merges explicit edges from tests/test-map.json.

node scripts/affected-tests.mjs                     # list L1 (pure) and L3 (rig) tests
node scripts/affected-tests.mjs --json              # machine-readable selection
node scripts/affected-tests.mjs --run               # run L1 with two workers, L3 serially
node scripts/affected-tests.mjs --tier L1 --run     # skip the slow rig group while iterating
node scripts/affected-tests.mjs --files src/shared/chat-activity.ts --run
node scripts/affected-tests.mjs --staged --run      # only what is staged

Rig tests (anything reaching tests/helpers/studio-rig.ts) are L3 and run with --test-concurrency=1. A change under src/harness-seed/** or src/game-template/** also selects the harness gate (harness-incidents and scoreboard), because rigs copy those folders instead of importing them.

Area and named files

  • Area ids with their docs and sources: docs/agent/knowledge-map.json; npm run review:context -- --area <id> lists the docs and checks for one area.
  • One or more files: npm test -- tests/conformance/<name>.test.ts [more files]. This avoids building the affected graph, so prefer it for red/green iteration.
  • One behaviour while iterating: npm test -- --test-name-pattern='<test name>' tests/conformance/<name>.test.ts.

Rules

  • Red first: run the new or changed test and see it fail for the right reason.
  • A selector that picks nothing for a behaviour change means a test or a tests/test-map.json edge is missing; add it rather than skipping verification.
  • On failure keep the log, decide whether product, expectation, fixture or environment is wrong, and rerun only after a specific change. Never raise deadlines to get green.
  • Full npm test / npm run verify only for broad changes or when asked. Full Node tests run the fast group with two workers, then rigs serially; rigs still run after fast failures.
  • Report which layers and files ran.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.