Skip to content

AGENTS.md

Latest commit

 

History

History
31 lines (22 loc) · 2.39 KB

AGENTS.md

File metadata and controls

31 lines (22 loc) · 2.39 KB
Raw

Agent instructions — TimShermanMusic

Cross-AI hard rules

The cross-AI hard rules that bind every agent on every surface are NOT duplicated here. They live in exactly one file: docs/cross-ai-rules.md in the web-jam-tools repository, which normally sits alongside this repository — ../web-jam-tools/docs/cross-ai-rules.md, and on Josh's laptop /home/joshua/WebJamApps/web-jam-tools/docs/cross-ai-rules.md.

Read that file before acting. If you cannot find it, STOP and say so — do not proceed without the rules and do not reconstruct them from memory or from this file.

Pull requests

Never merge to dev or main — Josh is the mandatory human reviewer. Open PRs with the shared script (~/WebJamApps/web-jam-tools/scripts/create-draft-pr.sh), never gh pr create directly. It always opens a draft PR based on dev. The script requires --author, --summary, --test-plan, and --test-evidence; empty or placeholder values are rejected. Bump the semver version in package.json once per PR, on the PR's first commit only.

PR body conventions (violations may be machine-rejected)

  • Summary: markdown bullet points, one change per bullet — never a run-on paragraph.
  • Test evidence: paste the REAL runner output verbatim (the lines showing pass/fail and test counts), inside a ``` fence — never a description like "all tests passed". If the output has scrolled out of view, re-run the test command and paste what it prints.
  • Test plan: exact commands and manual steps that exercise the change (start command, route/page, what to click, expected visible result) — a green test suite alone is not a plan.
  • Attribution: --author names the model actually doing the work. Antigravity/agy sessions are ALWAYS agy — Gemini 3.5 Flash (Medium) or (High) — never write any other Gemini model name (models misremember their own identity; use this exact string).

Memory & Security Audits

  • Snyk Failures & Resolution via npm audit fix: PR checks may report failure on security/snyk due to transitive dependency vulnerabilities (such as brace-expansion or postcss). Running npm audit fix updates package-lock.json with non-breaking patches to resolve these vulnerabilities. Always run npm test and npm run test:lint afterwards to verify the test suite remains 100% green before committing and pushing the updated package-lock.json to the PR branch.