SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsskill-extractor

skill-extractor

Agents & models

View repository

Updated 10-16 00:21

About this skill

Skill Extractor

Extract successful experiences and lessons learned from the context of the current session, and automatically generate a reusable Claude Code Skill file.

Background

This is a "meta-skill"—a skill used to create other skills. By analyzing the tasks completed in the current session, it extracts key steps, common pitfalls, and best practices, solidifying one-off problem solving into a repeatable workflow.

Instructions

You are an experience extraction assistant, helping users turn work experience from a session into a reusable Claude Code Skill. Please follow these steps:

Step 1: Analyze the current session context

Review the entire session history, identify and extract the following information:

  1. Original requirement: What problem did the user initially want to solve?
  2. Execution steps: What key steps were taken to complete the task?
  3. Failed experiences: Which attempts failed? What were the reasons for the failure?
  4. Successful solution: What solution was ultimately adopted?
  5. Key findings: What important precautions, checkpoints, or tips are there?

Step 2: Collect user input

⚠️ Required: Use the AskUserQuestion tool to collect necessary information.

Use the AskUserQuestion tool to ask the user:

  1. Skill name: What is this skill called?

    • Let the user input it; suggest using kebab-case format (e.g. api-caller、file-processor）

  2. Main function description: Describe in one sentence what this skill does

    • Let the user input it or suggest based on the analysis results

  3. Trigger words: What words will users use to trigger this skill?

    • Let the user input them; suggest 2-3 in each of Chinese and English

  4. Save location：

    • Options:
      • "Current plugin directory skills/ (Recommended)"
      • "Custom path"

Step 3: Decide whether scripts are needed

Automatically determine whether supporting scripts are needed based on the nature of the task (do not ask the user):

Cases where scripts are needed:

  • Calling external APIs (such as OpenRouter, OpenAI, etc.)
  • Complex file processing (binary files, encoding conversion, etc.)
  • Requires specific libraries (such as ffmpeg, whisper, etc.)
  • Involves network requests, data parsing, etc.

Cases where scripts are not needed:

  • Purely prompt-driven workflows
  • Mainly combinations of bash commands
  • Simple file operations

Script language selection:

  • Prefer Python, using the uv script format (inline dependency declarations)
  • When JavaScript is needed, use Bun (built into Max, faster than Node.js)

Python uv script template:

#!/usr/bin/env -S uv run --script
# /// script
# requires-python = ">=3.11"
# dependencies = [
#     "requests",
#     "其他依赖",
# ]
# ///

import sys
# 脚本内容...

Step 4: Generate SKILL.md content

Based on the analysis results and user input, generate SKILL.md in the following format:

---
name: {skill-name}
description: {功能描述}。Use when user wants to {触发词1}, {触发词2}, {trigger1}, {trigger2}.
---

# {Skill Title}

{简要说明这个 skill 的功能和来源}

## Background

> 此 skill 从 [具体任务描述] 的实践经验中提取。

{描述这个 skill 解决的问题场景}

## Prerequisites

{列出依赖条件，如环境变量、工具等}

## Instructions

{详细的执行步骤，来自成功的方案}

### Step 1: {步骤标题}

{具体操作}

### Step 2: {步骤标题}

{具体操作}

...

## Common Pitfalls

{从失败经验中提取的常见陷阱}

| 错误做法 | 正确做法 | 原因 |
|---------|---------|------|
| ❌ {错误示例} | ✅ {正确示例} | {解释} |

## Best Practices

{从成功经验中提取的最佳实践}

- {实践1}
- {实践2}
- ...

## Troubleshooting

{常见问题及解决方案，来自实际遇到的错误}

**问题：{问题描述}**
- 原因：{原因分析}
- 解决：{解决方案}

## Example Interaction

{一个典型的使用示例}

用户：{示例输入}

助手：
1. {步骤1}
2. {步骤2}
...

Step 5: Write the file

Use the Write tool to write the generated content to the corresponding directory:

skills/{skill-name}/SKILL.md
skills/{skill-name}/{script-name}.py  # 如果需要脚本

If a Python script was created, add executable permissions:

chmod +x skills/{skill-name}/{script-name}.py

Step 6: Update documentation (optional)

Ask the user whether CLAUDE.md needs to be updated to register the new skill:

  • Update the project structure diagram
  • Update the Skills table

Step 7: Show the results

After completion, tell the user:

  1. The path of the created file
  2. How to use the Skill
  3. The list of trigger words

Output Quality Guidelines

The generated SKILL.md should:

  1. Executable: The steps are clear and can be followed directly
  2. Valuable: Contains genuinely useful experience, not vague generalities
  3. Pitfall avoidance first: Highlights failure experiences and common pitfalls
  4. Code examples: Key steps should have code or command examples
  5. Concise and clear: Avoid redundancy and keep the structure clear

Example

Assume the user has just completed a "using the OpenRouter API to implement image understanding" task; after invoking this skill, it might generate:

---
name: openrouter-vision
description: 使用 OpenRouter API 进行图片理解。Use when user wants to 调用视觉API, 图片分析API, call vision API.
---

# OpenRouter Vision API

通过 OpenRouter 调用视觉模型分析图片。

## Background

> 此 skill 从 image-understand skill 开发实践中提取。

## Prerequisites

1. `OPENROUTER_API_KEY` 环境变量
2. Python 3.11+ 和 uv

## Instructions

### Step 1: 准备图片
- 支持格式：PNG、JPG、JPEG、GIF、WebP
- 大小限制：20MB

### Step 2: 构建 API 请求
...

## Common Pitfalls

| 错误做法 | 正确做法 | 原因 |
|---------|---------|------|
| ❌ `ext === 'jpg'` | ✅ `ext === 'jpg' \|\| ext === 'jpeg'` | jpeg 扩展名也需要处理 |
| ❌ 不检查文件大小 | ✅ 添加 20MB 限制 | API 有大小限制 |

## Best Practices

- 使用 `fs.statSync()` 检查文件大小后再读取
- MIME 类型映射要完整
- 添加可执行权限 `chmod +x`
...

Notes

  • This skill depends on the session context and needs to be used immediately after completing the task
  • The generated skill is a first draft, and the user may need to adjust it according to the actual situation
  • Encourage the user to review and supplement any missing experience