Skill router
⚡ Blazing-fast skill routing engine for AI coding agents. Discovers, indexes, and routes 300+ agent skills in <50ms. Supports Gemini/OpenAI embedding reranking.
npx -y skills add LittlePeter52012/skill-routerAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
ALWAYS LOADED. Intelligent skill search router. When no installed skill matches the user's request, or when the user explicitly asks to find/search for a skill, run: skrt query '<user request>'. Read the top result's SKILL.md and follow its instructions. 智能技能路由器,当找不到匹配技能时自动搜索。触发词: find skill, search skill, 查找技能, 搜索技能, skillmux, skrt.
SKILL.md
2.5 KB, as published. Nobody here has run it
SKRT — Skill Router 智能技能搜索路由器
This meta-skill enables automatic skill discovery from your entire skill library (300+ skills) in < 50ms.
When to Use
Use this skill when:
- The user's request doesn't match any currently loaded skill
- The user explicitly asks to find or search for a skill
- You need to discover which skill handles a specific task
How to Use
Step 1: Search
Run the following command with the user's request as the query:
skrt query "<user's request or task description>"
Examples:
skrt query "NotebookLM search"
skrt query "用NotebookLM查找资料"
skrt query "PDF merge files"
skrt query "write scientific paper"
skrt query "brainstorm ideas"
skrt query "molecular docking" --top 10
Step 2: Parse Results
The command outputs JSON with matched skills sorted by relevance:
{
"query": "...",
"elapsed_ms": 3.2,
"provider": "local",
"results": [
{
"rank": 1,
"name": "skill-name",
"score": 95,
"path": "/absolute/path/to/SKILL.md",
"summary": "Brief description..."
}
]
}
Step 3: Load the Best Match
- Take the top result's
pathfield - Read that SKILL.md file using
view_file - Follow the instructions in that SKILL.md to complete the user's task
Step 4: If No Results
If results is empty or scores are very low (< 30):
- The skill library doesn't have a matching skill
- Proceed with your general knowledge
- Suggest the user install a relevant skill
Configuration
# Pin high-priority skills (always boosted in results)
skrt pin add brainstorming
skrt pin add prompt-master
# Check status
skrt status
# Force index rebuild after installing new skills
skrt index --force
Notes
- The
skrtbinary is installed at~/go/bin/skrt - Config:
~/.skrt/config.json - Index cache:
~/.skrt/index.json - Supports Chinese (CJK), English, and mixed queries
- Typical latency: < 80ms end-to-end
- Codebase:
~/path/to/skill-router/(GitHub:skrt-dev/skill-router)