agentleFS
Sign inStart free
diegopachecodiegopacheco/agent-skills/claude-agents-md/AGENTS.md

agent-skills / claude-agents-md

AGENTS.md0 starsChanged 3 months agoUnlicenseView on GitHub

The file

Show its rules and text

We split this AGENTS.md into the rules it gives an agent

46 rules. About 0 are common, 42 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈42

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every 30s you will share progress x/y done and what you doing in 1-2 lines.

    line 3
  2. Distinctive:

    Never use comments, never comment anything.

    line 6
  3. Distinctive:

    Never use the words: "demo", "demonstration" or "example", never ever.

    line 7
  4. Distinctive:

    Make sure the code is as simple as possible.

    line 8
  5. Distinctive:

    Make sure the code is well written and make sense.

    line 9
  6. Distinctive:

    Do not do things I did not ask for in explicit prompts.

    line 10
  7. Distinctive:

    Use the least number of libraries as possible, if possible no libraries.

    line 11
  8. Distinctive:

    Never add Co-Authored-By lines to commits

    line 14
  9. Distinctive:

    Always explain the task

    line 17
  10. Distinctive:

    Always put the whole prompt on the PR description

    line 18
  11. Distinctive:

    Always run the tests and get the test output on the PR.

    line 19
  12. Distinctive:

    Never use comments, never ever.

    line 22
  13. Distinctive:

    Never do sleep bigger than 1

    line 23
  14. Distinctive:

    When need to wait for a POD or a docker container in docker or k8s make sure you will use a loop and check the condition and do max sleep 1.

    line 24
  15. Distinctive:

    Dont use icons and enomjis on bash script.

    line 25
  16. Distinctive:

    Prefer name the file as Containerfile

    line 28
  17. Distinctive:

    Never use Docker, always use podman and podman-compose

    line 29
  18. Distinctive:

    Make sure you are using the latest versions

    line 30
  19. Distinctive:

    Do not write comments, never comment.

    line 31
  20. Distinctive:

    Make sure you dont do ENTER between commands make it compact

    line 32
  21. Distinctive:

    Never use docker-compose always use podman-compose

    line 35
  22. Very specific:

    always have a start.sh where you start docker-compose.

    line 38
  23. Very specific:

    always have a stop.sh to stop docker-compose.

    line 39
  24. Very specific:

    always have a test.sh where you show the feature is working.

    line 40
  25. Distinctive:

    dont use timeouts more than 1 min

    line 41
  26. Distinctive:

    Never do ACII ART for architecture diagrams

    line 44
  27. Distinctive:

    Always do Excalidraw-style diagram.

    line 45
  28. Distinctive:

    When there is UI always use Playwright MCP to take print screens, add on printscreens/ folder on the project and always load, refer, render …

    line 46
  29. Distinctive:

    IF there is a logo, always display/render in the begining of the README

    line 47
  30. Distinctive:

    Always make sure it's just one version and one version only of the app installed

    line 57
  31. Distinctive:

    Make sure there are install/uninstall scripts

    line 58
  32. Distinctive:

    When making changes make sure remove, rebuild, re-install the app

    line 59
  33. Distinctive:

    Always light themed (never DARK - unless user asked for)

    line 62
  34. Very specific:

    IF does not have code, should be self contained (CSS/JS) inside single index.html (Not the case for apps with backend code)

    line 63
  35. Distinctive:

    Stop when confused. Name what's unclear.

    line 73
  36. Distinctive:

    No features beyond what was asked.

    line 77
  37. Distinctive:

    Test: would a senior engineer say this is overcomplicated?

    line 78
  38. Distinctive:

    Don't "improve" adjacent code, comments, or formatting.

    line 82
  39. Distinctive:

    Don't refactor what isn't broken.

    line 83
  40. Distinctive:

    Don't follow steps. Define success and iterate.

    line 87
  41. Distinctive:

    If two patterns contradict, pick one (more recent / more tested).

    line 91
  42. Distinctive:

    Don't blend conflicting patterns.

    line 93
  43. Distinctive:

    Before adding code, read exports, immediate callers, shared utilities.

    line 96
  44. Distinctive:

    Don't continue from a state you can't describe back.

    line 105
  45. Distinctive:

    If you lose track, stop and restate.

    line 106
  46. Distinctive:

    Default to surfacing uncertainty, not hiding it.

    line 111

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

CODEX Guidelines

  • For borwser use always do npx playwright
  • Every 30s you will share progress x/y done and what you doing in 1-2 lines.

General Guidelines

  • Never use comments, never comment anything.
  • Never use the words: "demo", "demonstration" or "example", never ever.
  • Make sure the code is as simple as possible.
  • Make sure the code is well written and make sense.
  • Do not do things I did not ask for in explicit prompts.
  • Use the least number of libraries as possible, if possible no libraries.

Git Commits

  • Never add Co-Authored-By lines to commits

Github Pull Request(PR) Guidelines

  • Always explain the task
  • Always put the whole prompt on the PR description
  • Always run the tests and get the test output on the PR.

When Writing BASH scripts

  • Never use comments, never ever.
  • Never do sleep bigger than 1
  • When need to wait for a POD or a docker container in docker or k8s make sure you will use a loop and check the condition and do max sleep 1.
  • Dont use icons and enomjis on bash script.

Containerfile

  • Prefer name the file as Containerfile
  • Never use Docker, always use podman and podman-compose
  • Make sure you are using the latest versions
  • Do not write comments, never comment.
  • Make sure you dont do ENTER between commands make it compact

podman-compose

  • Never use docker-compose always use podman-compose
  • This section only apply if you have podman-compose requirements otherwise ignore it.
  • docker-compose it's alias to podman, keep that in mind.
  • always have a start.sh where you start docker-compose.
  • always have a stop.sh to stop docker-compose.
  • always have a test.sh where you show the feature is working.
  • dont use timeouts more than 1 min

README.md

  • Never do ACII ART for architecture diagrams
  • Always do Excalidraw-style diagram. Rendered a hand-drawn flow diagram (wobble filter + "Caveat" handwriting font, pastel boxes, solid capture path.
  • When there is UI always use Playwright MCP to take print screens, add on printscreens/ folder on the project and always load, refer, render the images and explain on README.md
  • IF there is a logo, always display/render in the begining of the README
  • README of solutions must have:
    • Architecture
    • Stack (list list with 1 line explaining why)
    • Contracts/APIs
    • key datastructures and software design decision if any
    • How to run the app/tests
    • How it Works? (3-15 lines max)

Electron/Swift MacOS Apps

  • Always make sure it's just one version and one version only of the app installed
  • Make sure there are install/uninstall scripts
  • When making changes make sure remove, rebuild, re-install the app

Websites Design (index.html)

  • Always light themed (never DARK - unless user asked for)
  • IF does not have code, should be self contained (CSS/JS) inside single index.html (Not the case for apps with backend code)

CLAUDE.md — 12-rule template

These rules apply to every task in this project unless explicitly overridden. Bias: caution over speed on non-trivial work. Use judgment on trivial tasks.

Rule 1 — Think Before Coding

State assumptions explicitly. If uncertain, ask rather than guess. Present multiple interpretations when ambiguity exists. Push back when a simpler approach exists. Stop when confused. Name what's unclear.

Rule 2 — Simplicity First

Minimum code that solves the problem. Nothing speculative. No features beyond what was asked. No abstractions for single-use code. Test: would a senior engineer say this is overcomplicated? If yes, simplify.

Rule 3 — Surgical Changes

Touch only what you must. Clean up only your own mess. Don't "improve" adjacent code, comments, or formatting. Don't refactor what isn't broken. Match existing style.

Rule 4 — Goal-Driven Execution

Define success criteria. Loop until verified. Don't follow steps. Define success and iterate. Strong success criteria let you loop independently.

Rule 7 — Surface conflicts, don't average them

If two patterns contradict, pick one (more recent / more tested). Explain why. Flag the other for cleanup. Don't blend conflicting patterns.

Rule 8 — Read before you write

Before adding code, read exports, immediate callers, shared utilities. "Looks orthogonal" is dangerous. If unsure why code is structured a way, ask.

Rule 9 — Tests verify intent, not just behavior

Tests must encode WHY behavior matters, not just WHAT it does. A test that can't fail when business logic changes is wrong.

Rule 10 — Checkpoint after every significant step

Summarize what was done, what's verified, what's left. Don't continue from a state you can't describe back. If you lose track, stop and restate.

Rule 12 — Fail loud

"Completed" is wrong if anything was skipped silently. "Tests pass" is wrong if any were skipped. Default to surfacing uncertainty, not hiding it.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free