awesome-agent-skills
AGENTS.md18 starsChanged 5 days agoMITView on GitHub
When adding a new skill, you must add both language variants together: skills/en/<skill-name>/ and skills/zh-cn/<skill-name>-cn/. Single-language submissions are not accepted.
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 31,700 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 repository55 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- ax-extract-workflowSkill
- conversation-json-to-mdSkill
- feature-doc-splitterSkill
- frontend-quality-guardrailsSkill
- git-commit-onlySkill
- git-pushing-fastSkill
- git-push-secondary-merge-primarySkill
- git-stepwise-commitSkill
- git-weekly-reportSkill
- linkedinSkill
- logo-design-boardSkill
- mac-software-storage-cleanupSkill
- multi-lang-readmeSkill
- novel-writerSkill
- react-component-hook-guidelinesSkill
- routerbase-api-integrationSkill
- routerbase-media-generationSkill
- routerbase-model-routingSkill
- topic-bookmarks-reorganizerSkill
- tree-ring-memorySkill
- tweetclaw-twitter-automationSkill
- ui-layout-analyzerSkill
- wechat-theme-extractorSkill
- weixin-article-to-obsidianSkill
- xquik-dataSkill
- x-twitter-scraperSkill
- yuque-lakebook-exportSkill
- ax-extract-workflow-cnSkill
- conversation-json-to-md-cnSkill
- feature-doc-splitter-cnSkill
- frontend-quality-guardrails-cnSkill
- git-commit-only-cnSkill
- git-pushing-fast-cnSkill
- git-push-secondary-merge-primary-cnSkill
- git-stepwise-commit-cnSkill
- git-weekly-report-cnSkill
- linkedin-cnSkill
- logo-design-board-cnSkill
- mac-software-storage-cleanup-cnSkill
- multi-lang-readme-cnSkill
- novel-writer-cnSkill
- react-component-hook-guidelines-cnSkill
- routerbase-api-integration-cnSkill
- routerbase-media-generation-cnSkill
- routerbase-model-routing-cnSkill
- topic-bookmarks-reorganizer-cnSkill
- tree-ring-memory-cnSkill
- tweetclaw-twitter-automation-cnSkill
- ui-layout-analyzer-cnSkill
- wechat-theme-extractor-cnSkill
- weixin-article-to-obsidian-cnSkill
- xquik-data-cnSkill
- x-twitter-scraper-cnSkill
- yuque-lakebook-export-cnSkill
The file
Show its rules and text
We split this AGENTS.md into the rules it gives an agent
15 rules. About 0 are common, 8 are distinctive, and 7 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 58
When adding a new skill, you must add both language variants together: skills/en/<skill-name>/ and skills/zh-cn/<skill-name>-cn/.
- Distinctive: line 72
Use concise, instructional Markdown with clear headings.
- Very specific: line 78
Keep the folder name and SKILL.md frontmatter name aligned with the command name
- Distinctive: line 79
Every skill must have both English and Chinese variants, and the pair must stay aligned in purpose, structure, and trigger coverage
- Very specific: line 82
Treat frontmatter description as routing metadata for model skill selection.
- Very specific: line 85
Do not mention a specific agent in descriptions, such as Use when Codex is asked ....
- Very specific: line 87
Keep ## When to Use / ## 何时使用 as the concrete trigger checklist after ## Overview so validators and readers can find usage rules.
- Very specific: line 90
Keep bilingual docs separated (English default in README.md, Chinese in README.zh-CN.md)
- Distinctive: line 92
Prefer short sections, bullet lists, and copy-pastable examples.
- Distinctive: line 97
Verify links and paths in all edited Markdown files.
- Distinctive: line 99
If skill logic changes, install locally and run at least one prompt per trigger example.
- Very specific: line 100
Run both strict validators before committing: python3 scripts/validate_skills.py --strict and node scripts/validate-skills.js --strict.
- Distinctive: line 105
Follow Conventional Commits as seen in project history:
- Distinctive: line 117
Keep each PR focused on one logical change.
- Distinctive: line 121
When a skill is added or removed, update the following locations to keep the directory in sync:
≈ “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.
Files
Repository Guidelines
Project Structure & Module Organization
This repository is a skill pack for AI agents, organized by skill directories under skills/.
skills/en/: all English skillsskills/en/novel-writer/skills/en/multi-lang-readme/skills/en/conversation-json-to-md/skills/en/topic-bookmarks-reorganizer/skills/en/wechat-theme-extractor/skills/en/weixin-article-to-obsidian/skills/en/mac-software-storage-cleanup/skills/en/ui-layout-analyzer/skills/en/yuque-lakebook-export/skills/en/tweetclaw-twitter-automation/skills/en/git-weekly-report/skills/en/git-push-secondary-merge-primary/skills/en/git-pushing-fast/skills/en/git-stepwise-commit/skills/en/frontend-quality-guardrails/skills/en/react-component-hook-guidelines/skills/en/logo-design-board/skills/en/feature-doc-splitter/skills/en/x-twitter-scraper/skills/en/ax-extract-workflow/skills/en/linkedin/skills/en/tree-ring-memory/skills/en/xquik-data/
skills/zh-cn/: all Chinese skillsskills/zh-cn/novel-writer-cn/skills/zh-cn/multi-lang-readme-cn/skills/zh-cn/conversation-json-to-md-cn/skills/zh-cn/topic-bookmarks-reorganizer-cn/skills/zh-cn/wechat-theme-extractor-cn/skills/zh-cn/weixin-article-to-obsidian-cn/skills/zh-cn/mac-software-storage-cleanup-cn/skills/zh-cn/ui-layout-analyzer-cn/skills/zh-cn/yuque-lakebook-export-cn/skills/zh-cn/tweetclaw-twitter-automation-cn/skills/zh-cn/git-weekly-report-cn/skills/zh-cn/git-push-secondary-merge-primary-cn/skills/zh-cn/git-pushing-fast-cn/skills/zh-cn/git-stepwise-commit-cn/skills/zh-cn/frontend-quality-guardrails-cn/skills/zh-cn/react-component-hook-guidelines-cn/skills/zh-cn/logo-design-board-cn/skills/zh-cn/feature-doc-splitter-cn/skills/zh-cn/x-twitter-scraper-cn/skills/zh-cn/ax-extract-workflow-cn/skills/zh-cn/linkedin-cn/skills/zh-cn/tree-ring-memory-cn/skills/zh-cn/xquik-data-cn/
README.md: default English project overviewREADME.zh-CN.md: Chinese overviewLICENSE,.gitignore: repository-level metadata
When adding a new skill, you must add both language variants together: skills/en/<skill-name>/ and skills/zh-cn/<skill-name>-cn/. Single-language submissions are not accepted.
Build, Test, and Development Commands
This repository has no compile/build pipeline. Typical contributor commands are:
git status— inspect local changes before committinggit diff— review exact content changespython3 scripts/validate_skills.py --strict— validate skill structure, bilingual counterparts, and English skill language rulesnode scripts/validate-skills.js --strict— run the legacy/alternative strict validatorgit config core.hooksPath .githooks— enable the local pre-commit validation hookcp -r skills/ ~/.claude/skills/— install skills locally for manual validationfind skills -maxdepth 3 -type f— quick structure check before PR
Coding Style & Naming Conventions
Use concise, instructional Markdown with clear headings.
- File names:
kebab-casefor skill folders (for example,novel-writer-cn) - Skill command naming:
- English skills use the base command name with no
-ensuffix - Chinese skills append
-cn - Keep the folder name and
SKILL.mdfrontmatternamealigned with the command name - Every skill must have both English and Chinese variants, and the pair must stay aligned in purpose, structure, and trigger coverage
- English
skills/en/**/SKILL.mdfiles must not contain Chinese/Han characters; put Chinese examples, triggers, and localized wording in the matchingskills/zh-cn/**/SKILL.md. The only exception isskills/en/multi-lang-readme/SKILL.md, where language-switch examples may show native language names.
- English skills use the base command name with no
- Skill metadata and opening structure:
- Treat frontmatter
descriptionas routing metadata for model skill selection. Keep it short and focused on when the skill should be used. - English skill descriptions must start with
Use when .... - Chinese skill descriptions must use natural Chinese trigger wording, such as
当用户……时使用。; do not start Chinese descriptions with EnglishUse when. - Do not mention a specific agent in descriptions, such as
Use when Codex is asked ...; describe the user need or task instead. - Move longer purpose, capabilities, outputs, and boundaries into the first body section,
## Overview. - Keep
## When to Use/## 何时使用as the concrete trigger checklist after## Overviewso validators and readers can find usage rules.
- Treat frontmatter
- Required skill file:
SKILL.md - Optional files/folders (add only when needed):
README.md,scripts/,references/,assets/ - Keep bilingual docs separated (English default in
README.md, Chinese inREADME.zh-CN.md)
Prefer short sections, bullet lists, and copy-pastable examples.
Testing Guidelines
Testing is documentation-driven and manual:
- Verify links and paths in all edited Markdown files.
- Validate trigger examples are clear and actionable.
- If skill logic changes, install locally and run at least one prompt per trigger example.
- Run both strict validators before committing:
python3 scripts/validate_skills.py --strictandnode scripts/validate-skills.js --strict.
The .githooks/pre-commit hook runs strict skill validation, including the English skill language rule, when core.hooksPath points to .githooks.
Commit & Pull Request Guidelines
Follow Conventional Commits as seen in project history:
docs(readme): split bilingual docs with English as defaultchore(git): add .gitignore for macOS metadata
PRs should include:
- A short summary of what changed and why
- Affected paths (for example,
skills/en/novel-writer/SKILL.md) - Screenshots only when visual docs/rendering are relevant
- Linked issue/task if available
Keep each PR focused on one logical change.
Skill Directory Maintenance
When a skill is added or removed, update the following locations to keep the directory in sync:
- This file (
AGENTS.md) - Project Structure & Module Organization section README.md- Available Skills table and Trigger Examples sectionREADME.zh-CN.md- 可用技能 table and 触发示例 section- Badge count in
README.mdandREADME.zh-CN.md
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.

