How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.
addyosmaniAGENTS.mdaddyosmani/web-quality-skills/AGENTS.md
skills/*/; do npx skills-ref validate "$skill" || exit 1; done ``` ## Priority rules When reviewing code, prioritize by impact: 1. **Critical** — Security vulnerabilities, complete failures 2. **High** — Core Web Vitals failures
runkidsAGENTS.mdrunkids/skillshare/AGENTS.md
perform opportunistic cleanup. - **Keep the docs true.** When documentation disagrees with code, verify the code and fix the stale documentation in the same task. Update a topic whenever a change
MongoDBAGENTS.mdmongodb/mongo-java-driver/AGENTS.md
JAVA-XXXX` matching the Jira ticket (e.g., `JAVA-6143`) - **Commits:** Keep commits logical and reviewable — each commit should be a coherent unit of change - **TODO comments:** Must reference a Jira
MongoDBAGENTS.mdmongodb/mongo-java-driver/mongodb-crypt/AGENTS.md
code - Do not: modify the C API binding layer without understanding the libmongocrypt contract ## Key Packages - `com.mongodb.crypt.capi` — mongocryptd C API bindings (JNA) — **security-critical, do not modify without human review
BytePioneer-AIAGENTS.mdBytePioneer-AI/codex-host/AGENTS.md
planning | [feature-boundary-planner](.agents/skills/feature-boundary-planner/SKILL.md) | | Export changes, reference tracing and code deletion | [dep-refs](.agents/skills/dep-refs/SKILL.md) | | React implementation and performance review | [react-best-practices](.agents/skills/react-best-practices/SKILL.md) | | AI chat components and conversation
crbnosAGENTS.mdcrbnos/carbon/AGENTS.md
Repo audit → handoff plans | `.claude/skills/improve/SKILL.md` | | Review your own branch before PR | `.claude/skills/self-review/SKILL.md` | ## Core Principles - **Simplicity First:** Make every change as simple as possible. Minimize code impact. - **No Laziness:** Identify root
GoogleAGENTS.mdgoogle/zerocopy/zerocopy/AGENTS.md
unsafe-rust`](../skills/unsafe-rust/SKILL.md) skill. That skill is the source of truth for unsafe-code authoring and review methodology
felinicsAGENTS.mdfelinics/Memoh/AGENTS.md
totals, never their sum. - Automatic approval applies only to external PR workflow runs, not code review, merging, or publishing. Agents updating this guide must not remove or weaken these rules
Microsoft AzureAGENTS.mdAzure/azure-sdk-for-java/AGENTS.md
Contributing See [CONTRIBUTING.md](https://github.com/Azure/azure-sdk-for-java/blob/main/CONTRIBUTING.md) for detailed contribution guidelines, including: - Pull request requirements - Code review process - Testing requirements - Documentation standards - Version management ## Agent-Specific Instructions For detailed GitHub Copilot
NVIDIAAGENTS.mdNVIDIA/cccl/AGENTS.md
Cycles For a given task, you should: 1. Research. Search the web, read existing code, look up system/dependency headers / implementations of related functionality. Figure out best practices and common pitfalls
samanhappyAGENTS.mdsamanhappy/mcphub/AGENTS.md
source of truth. If this guide disagrees with the code, follow the code and update the guide; tests document expected behavior and should change only when behavior intentionally changes
VercelAGENTS.mdvercel/workflow/AGENTS.md
bundle is the VM code the route carries as an inline string, while the code hosting it sits in the framework output. **The gate does not cover the world adapters
SupabaseAGENTS.mdsupabase/cli/packages/config/AGENTS.md
compiled `dist/index.js` for a `browser` target and asserts the output excludes registry-only code, proving the package.json `sideEffects: false` claim against real compiled output rather than merely asserting it — plus
RAIT-09AGENTS.mdRAIT-09/obsidian-agent-client/AGENTS.md
Agent Client Plugin - LLM Developer Guide ## Overview Obsidian plugin for AI agent interaction (Claude Code, Codex, Gemini CLI, Mistral Vibe, OpenCode, Kiro, Hermes Agent, Pi, Grok Build, custom agents
Javis603AGENTS.mdJavis603/token-monitor/AGENTS.md
command like `ls src/shared/` over a hand-maintained one); verify claims against the code before writing them; delete anything that has gone stale — an outdated note is worse than none
apifyAGENTS.mdapify/agent-skills/agents/AGENTS.md
agent plugins (coding-agent skills+MCP bundles or OpenClaw/Hermes-style harnesses), AI framework packages (LangChain/LlamaIndex-style), or direct application clients via apify-client. Use when planning, creating, or reviewing an integration that
VercelAGENTS.mdvercel/chat/apps/docs/AGENTS.md
CopyPrompt` when a page should give readers a prompt they can copy into a coding agent. ## Routing and proxy guidelines - Keep App Router route files as thin adapters around package
VercelAGENTS.mdvercel/chat/packages/adapter-github/AGENTS.md
public type changed). Sample fixtures and AGENTS.md edits don't. ## Resources - [Ship a GitHub code review bot with Hono and Redis](https://vercel.com/kb/guide/ship-a-github-code-review-bot-with-hono-and-redis) — Walks through building a GitHub
QoderAIAGENTS.mdQoderAI/better-harness/AGENTS.md
requires them. ## Change Traceability Review - Use `.agents/skills/change-traceability-review/SKILL.md` as the Story/Spec/Test/Risk evidence-chain review guide, not a code-style guide. - Before review, merge, or commit of a non-trivial change
SupabaseAGENTS.mdsupabase/etl/AGENTS.md
every branch, commit, PR, and review comment as **public**. Never put real secrets, credentials, tokens, private URLs/hostnames, customer data, or production payloads in code, tests, docs, comments, examples, commits
An open format for instructions to coding agents, read by Codex, Cursor and others. Think of it as a README written for agents.
At the repository root, with more specific files in subdirectories. Agents read the one closest to the file they're editing.
Setup and test commands, code style, and the rules a new contributor would need to know.
Claude Code reads CLAUDE.md. A one-line CLAUDE.md that points at AGENTS.md covers both.
