My personal collection of specialized skills for AI agents, organized as discrete, reusable capabilities.
npx skills add definite-d/agent-skills| Skill | Description |
|---|---|
atomic-commits |
Create atomic, logically-grouped commits from uncommitted changes using dependency-aware analysis and ordering. |
granular-git |
Character-precise, non-destructive Git commits via Git's plumbing layer. |
state-machine-design |
Design software systems using the state machine pattern. |
Adapted from Claude's "Anatomy of a Skill". Each skill resides in its own subdirectory:
skills/
βββ <skill-name>/
βββ README.md # Human-readable documentation and examples
βββ SKILL.md (required)
β βββ YAML frontmatter (name, description required)
β βββ Markdown instructions
βββ Bundled Resources (optional)
βββ scripts/ - Executable code for deterministic/repetitive tasks
βββ references/ - Docs loaded into context as needed
βββ assets/ - Files used in output (templates, icons, fonts)
SKILL.mdβ Contains the skill's trigger conditions, procedures, and command references. Parsed by agents to determine when and how to use the skill.README.mdβ Explains the problem the skill solves, the solution approach, and quick-start examples.
- Create a subdirectory named after the skill (use kebab-case).
- Add
SKILL.mdwith proper YAML frontmatter (nameanddescription). - Add
README.mdwith context and examples. - Update the table above in this file in alphabetical order.