Skip to content

docs: add CLAUDE.md agent guidelines for express-openid-connect#859

Open
jd3vi1 wants to merge 2 commits into
masterfrom
claude-md-guidelines
Open

docs: add CLAUDE.md agent guidelines for express-openid-connect#859
jd3vi1 wants to merge 2 commits into
masterfrom
claude-md-guidelines

Conversation

@jd3vi1

@jd3vi1 jd3vi1 commented Jul 22, 2026

Copy link
Copy Markdown
Contributor

Summary

Adds AI-agent guidelines for express-openid-connect — a lean, guardrail-first CLAUDE.md plus a thin AGENTS.md that points to it, generated by the generating-claude-md skill and validated against the repo.

  • CLAUDE.md (157 lines) — persona, working principles, project overview/structure, three-tier Boundaries, Security, and one-line anchors into references/.
  • AGENTS.md — imports @./CLAUDE.md so non-Claude tools (Codex CLI, Gemini CLI) share the same guidance.
  • references/*.md — full detail offloaded on demand: commands, testing, code-style, git-workflow, pitfalls, docs-update.

What detection surfaced (express-openid-connect specific)

  • Express OIDC Relying Party middleware (CommonJS, openid-client v6, jose, joi) — persona reflects an SDK engineer treating the lib/config.js Joi schema and index.js exports as the public contract, with session/cookie and OIDC-flow correctness front of mind.
  • Auth0-Client telemetry header (base64 {name,version,env}, set in lib/client.js / lib/context.js, opt-out via enableTelemetry) → Always-Do boundary scoped to new outbound OIDC request paths through createCustomFetch, preserving the opt-out.
  • No live-tenant test tier — the default npm run test suite is unit-only (test/setup.js disables outbound network via nock); the slower end-to-end/ tier runs Puppeteer against a local oidc-provider with test-only creds, so there's no Ask-First credentialed-test rule.
  • Security posture — default response_type: id_token (Implicit + Form Post), PKCE/state/nonce on the code flow via openid-client, RS256 default (none rejected), encrypted httpOnly session cookies (lib/appSession.js + lib/crypto.js), secrets from env → in Security, not the persona.
  • Hand-written index.d.ts + .version — types and version sources that drift easily → keep-in-sync boundaries and a pitfall.
  • Code-to-docs mapping (library shape) — config options and exports mapped to README.md / EXAMPLES.md / examples/ / index.d.ts; CHANGELOG.md and the version-specific migration guides intentionally not tracked.
  • Prettier + ESLint (singleQuote, printWidth: 80, unix line endings) enforced via a pretty-quick pre-commit hook + CI lint → CI-enforced style guardrail.

Validation

Filesystem validation pass: every cited path, every package.json script / CI job, and every references/*.md pointer (6/6) verified to exist in the repo — no invented tooling.

Generated by the generating-claude-md skill.

Add AI-agent guidelines generated by the generating-claude-md skill:
a lean root CLAUDE.md (single source of truth), a thin AGENTS.md that
imports it for non-Claude tools, and six references/*.md offload files.
@jd3vi1
jd3vi1 force-pushed the claude-md-guidelines branch from 73b32c7 to e87b9e0 Compare July 22, 2026 11:24
@jd3vi1 jd3vi1 changed the title docs: add CLAUDE.md + AGENTS.md AI agent guidelines docs: add CLAUDE.md agent guidelines for express-openid-connect Jul 22, 2026
@jd3vi1
jd3vi1 marked this pull request as ready for review July 24, 2026 06:53
@jd3vi1
jd3vi1 requested a review from a team as a code owner July 24, 2026 06:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants