How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.
OpenAIAGENTS.mdopenai/codex-security/AGENTS.md
Keep protections for credentials, unsafe paths, and settings the user explicitly requests. - Prefer straightforward code and tests for real behavior. - Mention another `openai/` repository in comments or pull request descriptions
LangChainAGENTS.mdlangchain-ai/open-swe/AGENTS.md
time context, not required startup reading. - Treat source code and tests as authoritative. A brief's unknowns and review items are verification gaps, not automatic requirements. - Prefer the narrowest quiet
GoogleAGENTS.mdgoogle/adk-recipes/core/python/ambient-expense-agent/AGENTS.md
Workflow`**. Business rules stay in code (a `$100` threshold routes low-value expenses straight to auto-approval); only high-value expenses reach an LLM `review_agent`, which then **pauses
GoogleAGENTS.mdgoogle/adk-recipes/core/python/long-horizon-harness/AGENTS.md
Start-here** column names the exact real function to read — open it; that production code is the lesson. | Interface | Start-here (real function) | Transferable vs Horizon-specific | Deep-dive | |---|---|---|---| | Environment
omnigent-aiAGENTS.mdomnigent-ai/omnigent/AGENTS.md
Agent guidance Guidance for AI agents (Claude Code, Copilot, Cursor, etc.) working in this repository. See `CONTRIBUTING.md` for the full contributor workflow. ## Committing Run the `pre-commit` hook before committing
KuberwastakenAGENTS.mdKuberwastaken/claurst/docs/agents.md
definitions override built-in agents of the same name. ```json { "agents": { "review": { "description": "Senior code reviewer focused on correctness and security", "model": "anthropic/claude-opus-4-6", "temperature": 0.3, "prompt": "You are a senior
CloudflareAGENTS.mdcloudflare/cloudflare-os/AGENTS.md
kernel**: it defines the architecture and is held to a higher bar than UI/gatekeeper code. Reviewers read *every line* of `workshop-backend` and of API changes in `workshop-shared
GitHubAGENTS.mdgithub/codeql/unified/AGENTS.md
hand; regenerate it with `scripts/regenerate-node-types.sh` after changing the pinned swift-syntax version, then review the diff alongside the mapping in `extractor/src/languages/swift/swift.rs`. ## AST Mapping - The target AST shape is described
open-gsdAGENTS.mdopen-gsd/gsd-core/docs/AGENTS.md
agents** detailed in this section. Thirteen additional shipped agents (pattern-mapper, debug-session-manager, code-reviewer, code-fixer, ai-researcher, domain-researcher, eval-planner, eval-auditor, framework-selector, intel-updater
AWSAGENTS.mdawslabs/mcp/src/dynamodb-mcp-server/AGENTS.md
Commit with conventional commit format (commitizen is configured) 6. Submit pull request or create code review ### Commit Message Format Follow [Conventional Commits](https://www.conventionalcommits.org/): ``` [optional scope]: [optional body] [optional
NVIDIAAGENTS.mdNVIDIA/garak/AGENTS.md
thread. ### Accountability - Pure code-agent PRs are **not allowed**. A human submitter must understand and defend the change end-to-end. - The submitting human must review every changed line
ibelickAGENTS.mdibelick/ui-skills/AGENTS.md
headings. - Keep headings `font-medium` with `tracking-tight` where appropriate. - Use `JetBrains Mono` for code, commands, and monospace UI. - Keep labels and link text in sentence case. - Prefer light borders
openstatusHQAGENTS.mdopenstatusHQ/openstatus/AGENTS.md
reasoning from the code alone. - When debugging a failure, start from the incident or issue Polylane recorded: it carries the evidence an investigation already gathered. - Polylane reviews pull requests
MervinPraisonAGENTS.mdMervinPraison/PraisonAI/AGENTS.md
packages plus the TypeScript SDK: core protocols in `praisonaiagents/`, agentic terminal CLI in `praisonai-code/`, bots/gateway in `praisonai-bot/`, LLM fine-tuning + agent training in `praisonai-train/`, browser automation
backnotpropAGENTS.mdbacknotprop/plannotator/AGENTS.md
ExitPlanMode` via hooks, letting users approve or request changes with annotated feedback. Also provides code review for git diffs and annotation of arbitrary markdown files. > **Reusing the document UI (theme
CloudflareAGENTS.mdcloudflare/vinext/AGENTS.md
bash git clone --depth 1 --single-branch --branch canary https://github.com/vercel/next.js.git .nextjs-ref ``` --- ## Code Style & Dependencies ### Prefer Node.js Built-in APIs Always use Node.js built-in modules and APIs
GoogleAGENTS.mdgoogle/adk-go/AGENTS.md
session, or hand the diff to a subagent that did not write the code. An agent reviewing work it produced in the same session re-reads its own intent rather
NVIDIAAGENTS.mdNVIDIA/OpenShell/AGENTS.md
first. We design systems and use agents to implement them — this is not vibe coding. The product provides safe, sandboxed runtimes for autonomous AI agents, and the project itself
CloudflareAGENTS.mdcloudflare/workerd/AGENTS.md
upstream code. These patches are stored in the `patches/` directory and are applied during the build process. When updating these dependencies, it's important to review and update the corresponding
CloudflareAGENTS.mdcloudflare/workerd/src/workerd/api/streams/AGENTS.md
become dangling See `README.md` §Safety Pattern Catalog for detailed When/Why/How for each pattern. ## CODE REVIEW RULE When reviewing changes to streams code, check whether the change requires updates
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.
