Thank you for your interest in contributing!
- Use GitHub Issues
- Specify which skill, hook, or agent is affected
- Fork the repository
- Create a feature branch
- Make your changes
- Commit with a descriptive message
- Open a Pull Request
| Directory | What We Need | Guidelines |
|---|---|---|
skills/ |
Skill improvements, new skills | Follow SKILL.md format (YAML frontmatter + execution flow). Keep skills focused on one task. |
agents/ |
Agent improvements | Follow agent .md format (YAML frontmatter + role description). |
hooks/ |
Hook improvements, new event handlers | Test thoroughly — hooks run on every tool invocation. Avoid false positives. |
templates/ |
Template improvements | CLAUDE.md template should follow Two-Tier Context Strategy (10-15 high-impact rules only). |
checklist-templates/ |
New language checklists, improvements | Keep checklists actionable and verifiable. |
- Each skill lives in
skills/{name}/SKILL.md - Required frontmatter:
name,description,allowed-tools - Optional:
argument-hint,context,agent - Execution flow should be step-by-step with clear decision points
- Output format should use consistent markdown tables
- Hooks are defined in
hooks/hooks.json - Filter non-source files (node_modules, dist, images, etc.)
- Use
additionalContextfor suggestions, not blocking - Test with both TypeScript and Python projects
When adding a new feature to this plugin, consider adding the equivalent to:
- ai-dev-os-plugin-kiro (as steering rule)
- ai-dev-os-plugin-cursor (as .mdc rule)
- Translations are in
docs/i18n/{lang}/ - Currently supported:
ja,zh-CN,ko,es
Be respectful, constructive, and inclusive.
Languages: English | 日本語