菜单
Search for Skills on StudyClawHub. Use when a student says 'sch search', 'search StudyClawHub', 'find a skill on sch', 'browse StudyClawHub', or 'sch find'. LLM-driven semantic search — fetches all skills, understands user intent, ranks by relevance.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Create a new Skill from scratch. Use when a student says 'sch create', 'create a skill', 'new skill', '创建skill', or wants to start a fresh Skill project.
Delete a Skill or Agent from StudyClawHub. Use when a student says 'sch delete', 'delete skill', 'remove from StudyClawHub', 'unregister skill', or '删除skill'.
Check, add, or fix dependencies for a Skill or Agent. Use when a student says 'sch deps', 'check dependencies', 'install deps', 'add deps', 'fix deps', 'generate dependencies', or 'what do I need to run this skill'.
Install a Skill or Agent from StudyClawHub. Use when a student says 'sch install X', 'install from StudyClawHub', 'get skill from sch', or 'download from StudyClawHub'. Clones the repo and installs it locally.
Submit your Skill(s) to StudyClawHub. Use when a student says 'sch submit', 'submit to StudyClawHub', 'register on StudyClawHub', 'publish to sch', or 'sch publish'. Scans the repo for all SKILL.md files and registers them via GitHub Issues.
One-line summary. Include trigger phrases, e.g. Use when user says 'do X', 'run Y'.
| name | sch-search |
| description | Search for Skills on StudyClawHub. Use when a student says 'sch search', 'search StudyClawHub', 'find a skill on sch', 'browse StudyClawHub', or 'sch find'. LLM-driven semantic search — fetches all skills, understands user intent, ranks by relevance. |
| author | EnyanDai |
| version | 0.1.0 |
| tags | ["hub","search","discover"] |
| metadata | {"openclaw":{"requires":{"bins":["curl"]}}} |
You are helping a student search and discover Skills on StudyClawHub. You (the LLM) are the search engine — read all skills, understand what the student wants, and decide the best ranking yourself.
Ask the student what they're looking for. They might say:
If you already fetched skills-index.json earlier in this conversation,
reuse the cached result — do NOT fetch again. Unless the student explicitly
asks to refresh (e.g. "refresh", "重新拉", "latest", "最新的").
Otherwise, fetch it from:
https://trust-app-ai-lab.github.io/StudyClawHub/skills-index.json
Parse it as JSON. The skills array contains all registered skills with
fields: name, description, tags, author, version, stars,
installs, registeredAt, repo.
Read every skill's full information — name, description, and tags. Use your understanding of the student's query to rank them by relevance. Consider:
You decide the final order. There is no external ranking algorithm — you are the ranker.
Show the top results in a clear format:
1. [skill-name] by @author ★3 ↓12
description text here
Tags: tag1, tag2 | v1.0.0
2. [another-skill] by @author2 ★1 ↓5
...
Briefly explain why you ranked them this way if it's not obvious.
After showing results, offer the student:
/sch-install {name} to install a skill