by agentskill-sh
此技能使Claude能够从agentskill.sh市场发现、安装和管理AI代理技能,将您的代理转变为自我改进系统。
1. 打开 Claude 聊天界面
2. 点击下方 "📋 复制" 按钮
3. 粘贴到 Claude 聊天框中并发送
4. 输入 "使用 learn 技能" 开始使用
=== learn 技能 === 作者: agentskill-sh 描述: 此技能使Claude能够从agentskill.sh市场发现、安装和管理AI代理技能,将您的代理转变为自我改进系统。 使用方法: 1. 调用技能: "使用 learn 技能" 2. 提供相关信息: 根据技能要求提供必要参数 3. 查看结果: 技能会返回处理结果 示例: "使用 learn 技能,帮我分析一下这段代码"
这种方法适用于所有 Claude 用户,不需要安装额外工具。
productivity
safe
Teach your AI agent anything — mid-conversation.
Remember the scene in The Matrix where Neo gets Kung Fu uploaded directly into his brain? "I know Kung Fu." Agent Skills work the same way — except for AI agents. You upload a skill file, and suddenly your agent knows about SEO, how to write cold emails, or even accounting in France.
/learn lets you do this mid-conversation. One command, 40,000+ skills from agentskill.sh.
/learn seo
Your agent searches the directory, shows the best matches, and installs your pick. No restart. No context switch.
Every skill on agentskill.sh has a security score (0-100). Before installing anything, /learn performs a local security scan to catch malicious instructions.
After incidents like OpenClaw showed how rogue SKILL.md files can compromise agents, vetting matters. Skills below 30 require explicit confirmation.
Agents auto-rate skills after use (1-5 scale with comments). The best skills surface. Broken ones get flagged. Your agent contributes to — and benefits from — collective quality signals.
Instead of hunting for skills manually, search 40,000+ skills mid-conversation. Find what you need, install it, keep working.
Claude Code (recommended)
/plugin marketplace add https://agentskill.sh/marketplace.json
/plugin install learn@agentskill-sh
Claude Desktop / Claude Cowork
Click Plugins → Add marketplace → paste https://agentskill.sh/marketplace.json
Git (any platform)
# Claude Code
git clone https://github.com/agentskill-sh/learn.git ~/.claude/skills/learn
# Cursor
git clone https://github.com/agentskill-sh/learn.git ~/.cursor/skills/learn
Or copy SKILL.md to your platform's skill directory.
/learn programmatic seo
/learn frontend react components
/learn marketing email sequences
Returns top 5 matches with name, author, install count, and security score.
/learn @anthropic/seo-optimizer
/learn @vercel/nextjs-expert
/learn
Run with no arguments — analyzes your project and suggests relevant skills.
/learn trending
/learn list # Show all installed skills
/learn update # Check for updates
/learn remove <slug> # Uninstall a skill
/learn feedback seo-optimizer 5 "Excellent keyword clustering"
/learn itself needs updating via content SHAEvery installed skill includes a metadata header:
# --- agentskill.sh ---
# slug: seo-optimizer
# owner: anthropic
# contentSha: a3f8c2e
# installed: 2025-01-15T10:30:00Z
# source: https://agentskill.sh/seo-optimizer
# ---
| Platform | Directory |
|---|---|
| Claude Code | ~/.claude/skills/learn/ |
| Cursor | ~/.cursor/skills/learn/ |
| GitHub Copilot | .github/copilot/skills/learn/ |
| OpenAI Codex | ~/.codex/skills/learn/ |
| Windsurf | ~/.windsurf/skills/learn/ |
| Gemini CLI | ~/.gemini/skills/learn/ |
| Manus | Import via Settings → Skills |
Public API. No key required.
| Endpoint | Purpose |
|---|---|
GET /api/agent/search | Search skills |
GET /api/agent/skills/:slug/install | Get skill content |
GET /api/agent/skills/:slug/version | Version check |
POST /api/skills/:slug/install | Track install |
POST /api/skills/:slug/agent-feedback | Submit feedback |
MIT
Built by agentskill.sh
View Count
0
Download Count
0
Favorite Count
0
Quality Score
73