How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.
LumiwealthAGENTS.mdLumiwealth/lumibot/AGENTS.md
links. - Keep generated illustration code and invented trading results out of instructional examples. Runnable code and actual result evidence belong in the page itself. - Every generated image must be visually
rynfarAGENTS.mdrynfar/meridian/AGENTS.md
Meridian Meridian bridges supported coding clients to Claude through the Agent SDK and to Antigravity through the official agy CLI. ## Working scope Continue authorized implementation through relevant verification, corrections
rynfarAGENTS.mdrynfar/meridian/docs/agents.md
automatically. Meridian parses OMO's Task tool descriptions to extract subagent names (explore, code-review, etc.) and builds SDK AgentDefinitions so Claude can route to the correct agent. Internal orchestration
omnimind-aiAGENTS.mdomnimind-ai/OmniBot/AGENTS.md
prepare publishable code changes. - Codex-created issue fixes should use a bot branch and draft PR targeting the default branch, usually `main`; branch protection and maintainer review control the merge
yyjeqhcAGENTS.mdyyjeqhc/webcodex/AGENTS.md
Implement and review - User outcomes, scope, prohibitions, and acceptance conditions are hard constraints. Suggested files and steps are guidance unless explicitly mandatory; adapt them to verified code and conventions. - Follow
doobidooAGENTS.mddoobidoo/mcp-memory-service/.claude/directives/agents.md
Full automated review (5 iterations, safe fixes enabled) - Quality gate checks before review - Test generation for new code - Breaking change detection **Usage:** ```bash # Full automated review bash scripts/pr/auto_review.sh # Quality gate
WorldFlowAIAGENTS.mdWorldFlowAI/everything-claude-code/rules/agents.md
architect | System design | Architectural decisions | | tdd-guide | Test-driven development | New features, bug fixes | | code-reviewer | Code review | After writing code | | security-reviewer | Security analysis | Before commits | | build-error-resolver
xu-xiangAGENTS.mdxu-xiang/everything-claude-code-zh/AGENTS.md
用途 | 何时使用 | |-------|---------|-------------| | planner | 实现规划 | 复杂功能、重构 | | architect | 系统设计与可扩展性 | 架构决策 | | tdd-guide | 测试驱动开发 | 新功能、错误修复 | | code-reviewer | 代码质量与可维护性 | 编写/修改代码后 | | security-reviewer | 漏洞检测 | 提交前、敏感代码 | | build-error-resolver | 修复构建/类型错误 | 构建失败时
xu-xiangAGENTS.mdxu-xiang/everything-claude-code-zh/.codex/AGENTS.md
Workflow） - `agents/openai.yaml` — Codex 接口元数据 可用技能： - tdd-workflow — 测试驱动开发（TDD），包含 80% 以上的覆盖率 - security-review — 全面的安全自查清单 - coding-standards — 通用编码规范 - frontend-patterns — React/Next.js 模式 - frontend-slides — 视口安全（Viewport-safe）的 HTML 演示文稿及
xu-xiangAGENTS.mdxu-xiang/everything-claude-code-zh/rules/agents.md
planner | 实现规划 | 复杂特性、重构 | | architect | 系统设计 | 架构决策 | | tdd-guide | 测试驱动开发 (TDD) | 新特性、Bug 修复 | | code-reviewer | 代码审查 | 代码编写/修改后 | | security-reviewer | 安全分析 | 提交代码前 | | build-error-resolver | 修复构建错误 | 构建失败时 | | e2e-runner
SentryAGENTS.mdgetsentry/sentry-php/AGENTS.md
formatting, lint, and test commands for the code you changed. - Before handing back substantive code changes, run `composer check` when feasible and call out anything you could
the-open-engineAGENTS.mdthe-open-engine/zeroshot/AGENTS.md
files. Review instructions prohibit repairs and warn that other nodes may execute concurrently; this is an instructional boundary. Missing dependencies are reported with environment evidence rather than unrelated code repair
MicrosoftAGENTS.mdmicrosoft/vscode-mssql/AGENTS.md
layer. - `extension-toolkit/vscode` may depend on `base` and the VS Code API. - Production code must not import `extension-toolkit/vscode/testing`. - Import an explicit toolkit layer; do not import from
xenodiumAGENTS.mdxenodium/agent-shell/AGENTS.md
human-written summaries save the maintainer from having to parse lengthy generated text. - Review all code in your PR yourself and vouch for its quality. ## Contributing This is an Emacs
hanshuaikangAGENTS.mdhanshuaikang/nezha/AGENTS.md
而是退化到 `Math.max(MINIMUM_COLS=2, Math.floor(0/cell)) = 2`。放过 → `resize_pty` → `SIGWINCH` → Claude Code / Codex 等全屏 TUI 按 `cols=2` 重排，buffer **永久打散成一字一行不可恢复**。 - **后端 `resize_pty` 必须保留畸形尺寸兜底**——前端三层防御漏掉的话也得在这里挡住，防御纵深不允许只剩前端一层。 ### 禁止方向
NVIDIA-AI-BlueprintsAGENTS.mdNVIDIA-AI-Blueprints/video-search-and-summarization/.github/helm-sync/AGENTS.md
chart that retains an immutable `nvcr.io/nvstaging/vss-core/*` default, ignores either global override, or hard-codes a managed image in an umbrella profile as drift. - Continue to flag every other semantic
NVIDIA-AI-BlueprintsAGENTS.mdNVIDIA-AI-Blueprints/video-search-and-summarization/.github/skill-eval/AGENTS.md
links there, and (e) flag that this is an **auto-generated adapter for the reviewer to check**. The fork comment (`adapter-note-…md`) instead says the bot can't push
CloudflareAGENTS.mdcloudflare/workers-oauth-provider/AGENTS.md
never displaces `latest`. Both release workflows share the `npm-publish` concurrency group. ### Bonk (AI code review) Mention `/bonk` or `@ask-bonk` in PR comments to get AI-powered code review
ShenSeanChenAGENTS.mdShenSeanChen/waku-agent/AGENTS.md
brand file there, never copy one into `examples/`, and **do not move code between `hosted/` and `waku/` without saying which license it lands under.** 8. **Don't edit the copied
Microsoft AzureAGENTS.mdAzure/azure-sdk-for-go/AGENTS.md
Verify compatibility**: Ensure changes don't break backward compatibility without explicit approval ### For Code Reviews When assisting with code reviews: 1. Check compliance with [Azure Go SDK Guidelines](https://azure.github.io
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.
