agent-skills
AGENTS.md0 starsChanged 40 days agoMITView on GitHub
A collection of open-source agent skills for writing, editing, summarization, and business communications. Works seamlessly with Claude Code, Cursor, Codex, Windsurf, Antigravity, and any environment supporting the Agent Skills specification. Every skill follows the standard Agent Skills specification: Current version: 1.1.0. All future versions follow 1.x.x progression:
- Commits and pushes
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 31,439 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 repository5 other files it gives its agents.
- cold-email-writerSkill
- fewwords-summarizerSkill
- human-pencil-detSkill
- human-pencilSkill
- musk-email-writerSkill
The file
We split this AGENTS.md into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 43
Every skill follows the standard Agent Skills specification:
- Very specific: line 46
Must begin with valid YAML frontmatter containing name and description.
- Very specific: line 91
Verify GitHub Action status: gh run list --repo osspakistan/agent-skills
≈ “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
OSS Pakistan Agent Skills
A collection of open-source agent skills for writing, editing, summarization, and business communications. Works seamlessly with Claude Code, Cursor, Codex, Windsurf, Antigravity, and any environment supporting the Agent Skills specification.
Quick Commands & Verification
- Interactive skill selection:
npx skills add osspakistan/agent-skills(orbunx/pnpx) - Install all skills:
npx skills add osspakistan/agent-skills --all - Install individual skill:
npx skills add osspakistan/agent-skills --skill <skill-name> - Validate repository skills:
npx skills add . --list - Claude Code plugin marketplace:
- Add marketplace:
/plugin marketplace add osspakistan/agent-skills - Install plugin:
/plugin install skills@skills
- Add marketplace:
Project Structure
agent-skills/
├── .github/workflows/
│ └── release.yml # CI workflow for auto-packaging releases
├── .claude-plugin/
│ ├── marketplace.json # Claude Code marketplace configuration
│ └── plugin.json # Plugin manifest listing available skills
├── skills/ # Production skills directory
│ ├── fewwords-summarizer/ # Content summarization and briefing
│ ├── human-pencil/ # Natural prose editor & AI cliché remover
│ ├── human-pencil-det/ # DET writing practice & vocabulary bank
│ ├── musk-email-writer/ # Urgent, first-principles executive memos
│ └── cold-email-writer/ # High-converting cold outreach & investor pitches
├── LICENSE # MIT License
├── README.md # Human-facing documentation & install guides
└── AGENTS.md # Agent operating instructions & repository rulesSkill Architecture
Every skill follows the standard Agent Skills specification:
SKILL.md(Agent Instructions):- Must begin with valid YAML frontmatter containing
nameanddescription. - The
descriptionis critical: AI agents use it for intent detection and automatic activation. - Body contains actionable workflows, voice principles, and execution steps.
- Must begin with valid YAML frontmatter containing
README.md(Human Documentation):- Rendered on GitHub when browsing the skill folder.
- Contains: what the skill does, quick install commands, example prompts, and license.
references/(Optional Context & Modules):- Supplementary markdown files (e.g. classification trees, prompt templates, word banks).
- Referenced by relative links from
SKILL.md.
Git Workflows & Release Management
When to use git push vs git tag
| Action | Command | Purpose | When to Use | Triggers CI Release? |
| Initial Upstream Link | git push -u origin main | Sets upstream tracking for branch | Run once on repository creation or when pushing a newly created branch. | No |
| Routine Code Sync | git push origin main | Pushes daily commits and work | Use constantly during routine skill writing, testing, editing prompts, and docs. | No |
| Official Version Release | git tag v1.x.x<br>git push origin --tags | Creates immutable version checkpoint | Only when ready to publish an official release (v1.0.1, v1.1.0, etc.). | Yes (Builds ZIPs & publishes GitHub Release) |
Semantic Versioning Guide (MAJOR.MINOR.PATCH)
Current version: 1.1.0. All future versions follow 1.x.x progression:
| Digit | Scope | When to Increment | Examples |
MAJOR (X.0.0) | Breaking changes | Architecture revamps, removing/deprecating skills, breaking output interfaces. | 2.0.0 |
MINOR (1.X.0) | New features | Adding a brand new skill (e.g. skills/brand-writer), adding new reference classifiers or prompt templates (fully backward-compatible). | 1.1.0, 1.2.0 |
PATCH (1.0.X) | Bug fixes & refinements | Prompt wording refinements in SKILL.md, correcting typos, updating examples in references/, docs fixes. | 1.0.1, 1.0.2 |
GitHub Actions Release & Deployment Pipeline
Automated by .github/workflows/release.yml:
- Trigger: Push of any git tag matching
v*(e.g.v1.0.1,v1.1.0), or manual workflow dispatch. - Build: Packages each skill folder into individual
.zipfiles (with the root folder preserved for Claude.ai), plusall-skills.zip. - Publish: Creates GitHub Release, generates changelogs, and uploads the downloadable ZIP assets.
Agent Release Checklist (When publishing a version)
- Bump
"version"in.claude-plugin/plugin.json(e.g."1.1.0"). - Validate local skills:
npx skills add . --list. - Commit changes:
git commit -am "chore(release): bump version to 1.1.0" - Tag and push:
git tag v1.1.0 && git push origin main --tags - Verify GitHub Action status:
gh run list --repo osspakistan/agent-skills
PR Review & Contribution Checklist
A pull request is ready to merge when:
- Every new skill has its own folder under
skills/<skill-name>/. SKILL.mdexists with valid YAML frontmatter (namematching the folder name exactly).- A human-friendly
README.mdis provided in the skill folder. - All relative links to
references/are valid and tested. - Skill is added to
.claude-plugin/plugin.jsonand the rootREADME.mdlist. npx skills add . --listdiscovers the skill without errors.
Gotchas & Common Pitfalls
- Directory name matches frontmatter:
skills/<name>directory name must be identical to thename:field inSKILL.md. - Do not commit
dist/: ZIP archives are generated on the fly by CI and should remain gitignored. - Root folder in ZIPs: Claude.ai requires the ZIP file to contain
<skill-name>/SKILL.md(notSKILL.mdat zip root). - No external code runtime needed: Skills are declarative prompt and reference files—do not add npm build scripts or compiler steps unless an explicit helper CLI script is needed.
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.

