# Skillpin

> Route specialized work through verified AI agent skills managed by skillpin.

- Skill: `nbsp1221/skillpin` (Agent Skill)
- Install (CLI): `npx skillmds@latest add nbsp1221/skillpin`
- Raw SKILL.md: https://api.skillmd.com/api/skills/nbsp1221/skillpin/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: nbsp1221 (https://skillmd.com/u/nbsp1221)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/nbsp1221/skillpin

---

# skillpin

Before specialized work, extract a few short, specific terms for the task and run:

```bash
skillpin search <terms> --json
```

Do not pass the entire user request as the search query. If the user explicitly asks to use skillpin, search is mandatory.

Compare the returned candidates by name, description, matchedFields, matchedTerms, source, relevance, and integrity status.
Read a returned `managedPath` only when `integrityStatus` is `verified`.

A result with `integrityStatus` of `modified` may have changed since it was locked. Before reading its `managedPath`, tell the user that it is modified and request explicit approval to read that exact managed path. Do not read it unless the user explicitly approves. If approval is denied, consider verified candidates or continue without a routed skill.

Search results do not include skill body or file content. Do not disclose managed skill content unless you have read an eligible path.

