How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.
Google CloudAGENTS.mdGoogleCloudPlatform/genmedia-creative-studio/AGENTS.md
Request** is opened. - **Action:** The agent reviews the code changes (diff), looking for bugs, security issues, and style improvements. - **Output:** It posts review comments directly on the PR. - **Note
hoangnb24AGENTS.mdhoangnb24/repository-harness/AGENTS.md
choices remain open, stop before edits; configurable defaults are not authority. - For architecture, reliability, security, or quality invariant work, read `docs/patterns/encoding-invariants.md` and enforce only accepted rules. - Report reusable agent friction
modu-aiAGENTS.mdmodu-ai/moai-adk/AGENTS.md
used to drop input validation at trust boundaries, error handling that prevents data loss, security measures, accessibility, or one runnable check behind non-trivial logic. If an implementation exceeds
i365devAGENTS.mdi365dev/free4chat/AGENTS.md
with a previously valid sessionId/participantId. Do not rely on `participantHandle`/sessionId opacity as a security boundary. - **Stop is real revocation, not just a UI toggle.** Stopping Live Transcript or revoking
Ido-LeviAGENTS.mdIdo-Levi/Hephaestus/AGENTS.md
type-check` before requesting review, calling out any skipped checks with rationale. ## Security & Configuration Tips - Store secrets in `.env`; use `hephaestus_config.yaml` or `config/agent_config.yaml` for overrides and never commit credentials. - Reset
matthiasnAGENTS.mdmatthiasn/lotti/AGENTS.md
command are in [knowledge/conventions/screenshots.md](knowledge/conventions/screenshots.md). - Update docs and localization as needed (run `make l10n`). ## Security & Configuration Tips - Never commit secrets. Use `.env` for local config; keep
affaan-mAGENTS.mdaffaan-m/agentshield/.codex/AGENTS.md
Playwright, and Sequential Thinking. ## Multi-Agent Support - Explorer: read-only evidence gathering - Reviewer: correctness, security, and regression review - Docs researcher: API and release-note verification ## Workflow Files - No dedicated workflow
Microsoft AzureAGENTS.mdAzure/GPT-RAG/AGENTS.md
docs` branch and at https://azure.github.io/GPT-RAG/. ## Priority Follow, in this order: 1. Security, privacy, authorization, and platform instructions. 2. Task requirements and acceptance criteria. 3. Executable configuration and versioned
keinsaasforeverAGENTS.mdkeinsaasforever/better-chatbot/AGENTS.md
coverage and manual steps. - Before opening PR: `pnpm check` (lint+types+tests) should pass. ## Security & Configuration Tips - Copy `.env.example` to `.env`; never commit secrets. For local HTTP use `NO_HTTPS
oceanbaseAGENTS.mdoceanbase/powercontext/AGENTS.md
breaking changes, list validation commands, and include the AI usage statement requested by `.github/pull_request_template.md`. ## Security & Configuration Tips Do not commit secrets, local virtual environments, caches, `.powercontext/`, `dist/`, or generated website
LING71671AGENTS.mdLING71671/open-reverselab/AGENTS.md
make_crypto_replay_scaffold`）而非手动操作 5. **按攻击网多路径推进** — crypto→protocol→cheating→firmware→hardware→radio→ai-security，领域交叉是常态 ## 目录约定 项目根目录是当前工作区。 推荐目录结构： - `tools/`：逆向工具 - `samples/`：待分析样本 - `projects/`：Ghidra、IDA、调试器项目文件 - `notes/`：分析笔记 - `scripts
tintinwebAGENTS.mdtintinweb/pi-subagents/AGENTS.md
entries go under `## [Unreleased]`, in the right subsection (`### Added`, `### Changed`, `### Fixed`, `### Removed`, `### Security`, `### Refactored`). Read the section first and append to existing subsections; never duplicate them. - One bullet per issue/PR
CloudflareAGENTS.mdcloudflare/sandbox-sdk/AGENTS.md
trying to fetch GitHub URLs directly. ## Project Overview The Cloudflare Sandbox SDK enables secure, isolated code execution in containers running on Cloudflare. The SDK allows Workers to execute arbitrary commands
CloudflareAGENTS.mdcloudflare/sandbox-sdk/packages/sandbox/src/bridge/AGENTS.md
warm pool management, and authentication — all via Hono. Consumer-facing documentation (API reference, deployment, security) lives in [`bridge/worker/README.md`](../../../../bridge/worker/README.md). ## Key files - `index.ts` — `bridge()` factory: resolves DO bindings at module evaluation
MicrosoftAGENTS.mdmicrosoft/mssql-jdbc/AGENTS.md
Test-First Development**: All changes require tests — write failing tests before implementing fixes 5. **Security by Default**: Secure defaults, no credential logging, parameterized queries 6. **Protocol Compliance**: Follow
ClawBioAGENTS.mdClawBio/ClawBio/AGENTS.md
SKILL.md methodology or cited databases (CPIC, PharmGKB, ClinVar, etc.). Never invent bioinformatics values. 4. **Security filtering**: `clawbio.py` enforces per-skill `allowed_extra_flags` whitelists (INT-001). Do not bypass this
tomasz-tomczykAGENTS.mdtomasz-tomczyk/crit/AGENTS.md
stale_review_days` are **global config only**; project-level config cannot override (security — prevents malicious repos from hijacking the agent command, redirecting share requests to an attacker-controlled host, weakening
VercelAGENTS.mdvercel-labs/eve-software-factory-template/AGENTS.md
access — station instructions especially, since stations see none of the root's surface. ## Security - **Never ask the user for API keys, client secrets, or any other credentials.** - **Never commit secrets
TencentCloudBaseAGENTS.mdTencentCloudBase/CloudBase-AI-Toolkit/AGENTS.md
需求/设计/任务文档（**本地目录，不提交到公开仓库**。含内部信息，仅保留 `plugin-marketplace-listing/` 与 `npm-supply-chain-security-hardening/` 两个被代码或规范依赖的子目录；完整历史归档在私有仓库 `cloudbase-mcp-specs-archive`） 2. AGENTS 文件约定 - `AGENTS.md` 为项目及子目录的唯一可信源 - `CLAUDE.md`、`CODEBUDDY.md` 均为指向
withkynamAGENTS.mdwithkynam/vibecode-pro-max-kit/AGENTS.md
agent-browser` - `vc-web-testing` - `vc-frontend-design` - `vc-predict` - `vc-scenario` - `vc-security` - `vc-autoresearch` - `vc-debug` - `vc-agent-strategy-compare` - `vc-intent-clarify` - `vc-autopilot` Former workflow-owner
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.
