AGENTS.md examples from real projects

How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.

Best matches · from page 12Worked for most · soon

strands-agentsAGENTS.md

harness-sdk / strands-py

strands-agents/harness-sdk/strands-py/AGENTS.md

means exactly that: no tool catches it, so it is on you and the reviewer. ## Coding Patterns and Best Practices ### Logging Structured logging is a cross-SDK rule

8.5k26d agoDiscuss

strands-agentsAGENTS.md

harness-sdk / strands-ts

strands-agents/harness-sdk/strands-ts/AGENTS.md

reviewer. ### 5. Testing Guidelines When writing tests, you **MUST** follow the guidelines in [docs/TESTING.md](./docs/TESTING.md); see [Testing](#testing) below. ## Coding Patterns and Best Practices ### Logging Structured logging is a cross

8.5k26d agoReads credentialsDiscuss

strands-agentsAGENTS.md

harness-sdk / test-infra

strands-agents/harness-sdk/test-infra/AGENTS.md

package) - Running most integration tests (they don't need provisioned infra) - Reviewing or modifying SDK code, docs, tools, or CLI - Opening a PR — CI runs infrastructure-dependent tests automatically against

8.5k26d agoDiscuss

max-sixtyAGENTS.md

worktrunk

max-sixty/worktrunk/AGENTS.md

cargo run -- hook pre-merge --yes for the project test and lint gate. Claude Code web first runs task setup-web; Codex Cloud uses .codex/cloud.sh. Test and coverage recipes live

8.4k5mo agoDiscuss

max-sixtyAGENTS.md

worktrunk

max-sixty/worktrunk/docs/AGENTS.md

checks built links. Use root-relative links and asset paths. Starlight owns navigation, search, code frames, copy controls, and theme selection; keep overrides narrow. The design uses warm paper, dark

8.4k5mo agoDiscuss

maximhqAGENTS.md

bifrost

maximhq/bifrost/AGENTS.md

# AGENTS.md — Bifrost AI Gateway > Context for AI agents (Claude Code, Copilot, Cursor, etc.) working

8.4k4d agoReads credentialsDiscuss

OpenCoworkAIAGENTS.md

open-codesign

OpenCoworkAI/open-codesign/AGENTS.md

another preset. ### Built-In Starters And Scaffolds - Treat starter/scaffold assets as product code, not prompt filler. Review the actual file contents as well as the manifest before changing or adding

8k5mo agoPipes a download into a shellDiscuss

logancyangAGENTS.md

obsidian-copilot

logancyang/obsidian-copilot/AGENTS.md

running a multi-step dev session | [`designdocs/agents/PROCESS_GUIDE.md`](./designdocs/agents/PROCESS_GUIDE.md) | | changing user-facing behavior | [`designdocs/agents/DOCS_GUIDE.md`](./designdocs/agents/DOCS_GUIDE.md) | | reviewing code or preparing an Obsidian submission | [`designdocs/OBSIDIAN_COMMUNITY_REVIEW.md`](./designdocs/OBSIDIAN_COMMUNITY_REVIEW.md) | ## Important notes - The plugin supports multiple

7.8k8d agoDiscuss

VercelAGENTS.md

native

vercel-labs/native/AGENTS.md

Cmd.request`; do not import a service from the core. Do not add Zig app code unless the user explicitly chooses `--template zig-core` or the feature requires a toolkit extension

7.7k44d agoDiscuss

Gentleman-ProgrammingAGENTS.md

gentle-ai

Gentleman-Programming/gentle-ai/AGENTS.md

Index When working on this project, load the relevant skill(s) BEFORE writing any code. Naming convention: `gentle-ai-*` skills are repo-specific workflow skills. Unprefixed skills are portable writing

7.4k7mo agoDiscuss

Gentleman-ProgrammingAGENTS.md

gentle-ai

Gentleman-Programming/gentle-ai/docs/agents.md

7.4k7mo agoDiscuss

trailofbitsAGENTS.md

skills

trailofbits/skills/AGENTS.md

guarantee: - **the two loadability checks**, which need the Claude Code and Codex CLIs installed - **review-walkthrough's browser checks**, which need Chrome or Chromium. Run `node plugins/review-walkthrough/tests/browser-check.mjs /path/to/chromium` locally

7.3k9mo agoDiscuss

NVIDIAAGENTS.md

warp

NVIDIA/warp/AGENTS.md

files. Rebuild native libraries only after changes to `warp/native/` C++/CUDA code: - Standard build: `uv run build_lib.py` (~5 min) - Quick build: `uv run build_lib.py --quick` (~2-4 min) - Only

7.2k40d agoDiscuss

tw93AGENTS.md

Waza

tw93/Waza/AGENTS.md

constraints should be extracted from public repository context or user-provided task context. - Treat `code-review` as an invocation alias for Waza `check`, not as a separate generic skill. - Skill

7.1k4mo agoDiscuss

MicrosoftAGENTS.md

Security-101

microsoft/Security-101/AGENTS.md

requires PR reviews - All changes go through pull request process - Translation files are auto-generated, do not edit manually - Use meaningful commit messages ## Notes for AI Coding Agents - **This