My own collection of agent skills
- Python 95.5%
- Just 2.6%
- Dockerfile 1.9%
| .agents/skills/writing-great-skills | ||
| skills/search-engine | ||
| src/benchmark | ||
| .gitignore | ||
| justfile | ||
| README.md | ||
| skills-lock.json | ||
Agent Skills
My own set of skills for AI agent.
Skills
| Skill | Description |
|---|---|
| search-engine | Web search via SearXNG |
Installing skills
bunx skills add https://git.napatsc.com/ns/skills.git
See each skill's SKILL.md for configuration and usage.
Benchmarking
Since I am not sure that my additional skills would improve the agent or not, I also has the skill benchmark section. It measures whether the skill inclusion actually improve agent performance: fewer tokens, better answers, or faster completion.
cd benchmark && uv run benchmark