[luinbytes]

Why

Coding CLIs read AGENTS.md and CLAUDE.md at the start of every run. This repo had neither. These files record the real PR checks, layout, the master default branch, and gotchas, including that timing-sensitive tests must use a handshake rather than sleeps, retries, or longer timeouts.

What changed

• Added AGENTS.md with Node 22 setup, npm ci, npm audit --audit-level=high, npm run lint, npm run test:pond, and PATH="$PWD/.venv/bin:$PATH" npm run test:e2e (which already runs npm run build). Notes that pushes to master deploy via GitHub Pages after checks.
• Added CLAUDE.md containing @AGENTS.md.

No application, test, or workflow changes. The verify-luinbytes-dev skill is already in open PR #97 and is not part of this PR.

Checks

On the coding box: npm ci, npm audit --audit-level=high (0 vulnerabilities), npm run lint, and npm run test:pond (54 passed, 0 failed). test:e2e was not run locally (Playwright browser suite not set up in this pass). PR CI (checks.yml) should run the full gate, including e2e.

---

Note

Low Risk
Documentation-only; no runtime or CI logic changes.

Overview
Adds AGENTS.md so coding CLIs have a single source of truth for this repo: layout, Node 22 + local Python/Playwright setup (including CI vs local Playwright install rules), branch/workflow rules, required gate commands (npm audit, lint, pond tests, e2e with build), dev commands, sitemap revert gotcha, and master deploy behavior.

Adds CLAUDE.md as a one-line shim (@AGENTS.md) so Claude-oriented tools load the same guidance. No application, test, or workflow code changes.

^{Reviewed by Cursor Bugbot for commit ed03f61. Bugbot is set up for automated code reviews on this repo. Configure here.}