skills
AGENTS.md1 starsChanged 4 days agoMITView on GitHub
Installable skills live in skills/productivity/, skills/engineering/, and skills/creative/. Each kebab-case skill directory has a SKILL.md entrypoint and an agents/openai.yaml metadata file; add scripts/, references/, or local fixtures only when the skill needs them. Python regression suites are grouped by product under tests/<skill-name>/. docs/archive/ preserves upstream or historical documentation, while README.md and README_CN.md are the current catalog. Keep license provenance in LICENSES/. This repository has no build step or local package manifest. Run checks from the repository root: The first command…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 32,671 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository25 other files it gives its agents.
- game-experience-reviewSkill
- image-quality-controlSkill
- reference-image-promptSkill
- suno-createSkill
- suno-music-explorerSkill
- suno-studio-downloadSkill
- consensus-change-reviewSkill
- consensus-review-loopSkill
- delegated-change-reviewSkill
- docs-context-curatorSkill
- godot-silent-reviewSkill
- implement-in-stagesSkill
- improve-code-commentsSkill
- review-gated-implementationSkill
- review-loopSkill
- review-testsSkill
- codex-session-recoverySkill
- codex-thread-reportSkill
- consensus-gated-grillingSkill
- grillingSkill
- opencode-session-toolkitSkill
- review-gated-grillingSkill
- sanitize-artifactsSkill
- skill-manual-onlySkill
- wait-whatSkill
The file
We split this AGENTS.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 25
Use kebab-case for skill directories and keep the frontmatter name identical to the directory name.
≈ “Common” is an estimate: at least 1 in 5 of the 41,581 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.
Files
Repository Guidelines
Project Structure & Module Organization
Installable skills live in skills/productivity/, skills/engineering/, and skills/creative/. Each kebab-case skill directory has a SKILL.md entrypoint and an agents/openai.yaml metadata file; add scripts/, references/, or local fixtures only when the skill needs them. Python regression suites are grouped by product under tests/<skill-name>/. docs/archive/ preserves upstream or historical documentation, while README.md and README_CN.md are the current catalog. Keep license provenance in LICENSES/.
Build, Test, and Development Commands
This repository has no build step or local package manifest. Run checks from the repository root:
NO_COLOR=1 npx -y skills@latest add . --list
python3 -m unittest discover -s tests/repository-contract -p 'test_*.py' -v
python3 -m unittest discover -s tests/docs-context-curator -p 'test_*.py' -v
python3 -m unittest discover -s tests/codex-session-recovery -p 'test_*.py' -v
python3 -m unittest discover -s tests/opencode-session-toolkit -p 'test_*.py' -v
python3 -m unittest discover -s tests/suno-music-explorer -p 'test_*.py' -v
git diff --checkThe first command verifies catalog discovery. The Python commands check repository contracts and the scripted skill products. git diff --check catches whitespace errors. CI runs the same checks on pushes and pull requests.
Coding Style & Naming Conventions
Use kebab-case for skill directories and keep the frontmatter name identical to the directory name. All current skills are manual-only: pair disable-model-invocation: true in SKILL.md with policy.allow_implicit_invocation: false in agents/openai.yaml. Keep paired variants described in the root README behaviorally parallel except for their stated difference.
Assume the model already knows common domain knowledge and practices. Keep only instructions that change behavior for the task: its workflow, decisions, safety limits, tool contracts, and output requirements. Arrange sections in task order and avoid duplicated rules or unnecessary mandatory steps and reads. Use short, familiar wording; avoid repeating general advice or inventing terms. Put rarely needed details in references/ and read them only when relevant; put repeated, fragile operations in scripts/.
Python targets 3.11 and the standard library. No repository-wide formatter is configured, so preserve nearby formatting.
Testing Guidelines
Tests use unittest. Keep fixtures deterministic and independent of real user databases or home directories. There is no numeric coverage gate; add focused regression tests for changed scripts and repository-contract tests for packaging rules.
Commit & Pull Request Guidelines
Recent history uses short Conventional Commit-style subjects such as feat:, docs:, refactor:, chore:, and ci:. Use an imperative, lowercase summary and keep each commit to one intent. Pull requests should identify affected skills, explain catalog or paired-variant impact, list validation commands and results, and link an issue when applicable. Include screenshots only for visual documentation changes.
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
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.

