Skip to content

Agent Skills

Collection of skills for agent systems. Each skill provides specific capabilities that agents can discover, load, and execute.

Skills

Skill Description
agentctl Validate and generate hub indexes, manage hub registry, install/update/remove skills, manage local config
research-paper Process academic papers (PDF or URL) into a knowledge base via extract → analysis → synthesis pipeline
marker-pdf Convert PDFs, images, PPTX, DOCX, XLSX, HTML, EPUB to markdown, JSON, or HTML with OCR and table support
python-scaffold Scaffold a new Python project with modern toolchain (uv, ruff, pyright, pytest)
python-quality Run Python code quality checks — formatting, linting, type checking, and tests
semantic-commit Create semantic commits with interactive prompts following sp_theory standards
github-pages-setup Set up GitHub Pages with MkDocs Material and automatic deployment workflow
asdf Manage multiple runtime versions (Python, Node.js, Go, etc.) per project
jq Query, filter, transform, and format JSON from the command line

Skill Development

Skills follow the agent-foundation skills specification:

  • SKILL.md with machine-readable YAML frontmatter
  • lifecycle.yaml for install/update/uninstall
  • scripts/ for helper scripts (optional)
  • assets/ for Python/code examples (optional)
  • references/ for detailed documentation (optional)
  • Use SKILL_PATH variable provided by agent

For complete authoring guidelines, see agent-foundation skills authoring guide.

Organization

  • Each skill in its own directory
  • Named descriptively (e.g., marker-pdf, asdf)
  • Self-contained with lifecycle.yaml
  • Compatible with agent skill discovery systems
  • Minimal structure - only include necessary files

License

MIT

Contributing

See CONTRIBUTING.md.