find-my-skills（找到我的技能）

Find My Skills - Local Skills Intelligent Recommendation System

An intelligent local skills recommendation system that helps you quickly find the most suitable tools from your installed skills.

✨ Core Features

1. Local-First Search: Focus on your installed skills, avoid duplicate installations
2. Context-Aware: Precise recommendations based on conversation history and project progress
3. Multi-Dimensional Classification: Workflow stages, professional domains, function types, and more
4. Flexible Classification Strategies: 5 preset strategies plus custom strategies
5. Automatic Index Generation: One-click scan of local skills and index creation

🚀 Quick Start

First-Time Use

If you're using find-my-skills for the first time, you need to initialize:

/find-my-skills:init

The initialization process will:

1. Scan your local ~/.claude/skills/ and ~/.agents/skills/ directories
2. Extract metadata (name, description) from each skill
3. Let you choose a classification strategy (workflow-oriented, domain-oriented, functional-oriented, etc.)
4. Generate a personalized skills-catalog.md directory document
5. Create index data for quick recommendations

Estimated Time: 2-3 minutes (depending on the number of installed skills)

Choose Classification Strategy

During initialization, 5 classification strategies are provided:

💡 Recommendation: If unsure, choose "Domain-Oriented" or "Workflow-Oriented"

📌 When to Use find-my-skills vs find-skills

Use /find-my-skills (this skill) when:

• ✅ Need to recommend the most suitable tools from locally installed skills
• ✅ Want recommendations that consider project context and current progress
• ✅ Need skills for specific workflow stages (planning, implementation, review, etc.)
• ✅ Need skills for specific professional domains (bioinformatics, ML, etc.)
• ✅ Need skill combination suggestions and follow-up recommendations

Use /find-skills (external search) when:

• 🔍 Need to search online registry for new skills
• 📦 Need to install skills not yet installed
• 🆕 Want to explore latest community contributions
• 🌐 Unsure if a certain capability exists locally

Quick Judgment: If it contains words like "online", "search", "install", "new", use external find-skills; otherwise use this skill.

🎯 Usage Methods

Method 1: Direct Need Description (Recommended)

The most natural way, directly tell AI your needs:

"I need to analyze single-cell sequencing data"
"How to plan a new project"
"Is there a tool to generate PDF reports"

AI will:

• Analyze your conversation history and project context
• Understand your current work stage
• Recommend the most suitable skill combinations
• Explain recommendation reasons
• Provide usage examples

Method 2: Use Subskill Skills (Quick Navigation)

If you clearly know the type of need, you can directly use the corresponding subskill:

By Workflow Stage

By Professional Domain

By Function Type

🧠 Recommendation Logic

Context-First Principle ⚠️

Recommended skills must match the current project state and progress, not answer questions in isolation.

AI will:

1. Review conversation history: Understand current project background, goals, and principles
2. Assess project progress: Identify the stage the user is currently at
3. Analyze project characteristics: Understand tech stack, domain characteristics, scale and complexity
4. Consider completed work: Identify completed work and problems to solve
5. Understand user intent: Infer real needs from context

Multi-Dimensional Matching

The recommendation algorithm considers 4 dimensions (sorted by weight):

1. Workflow Stage (40%): Planning, implementation, review, documentation, deployment
2. Professional Domain (30%): Bioinformatics, machine learning, software engineering, etc.
3. File Type (20%):.docx,.py,.md, biological data formats, etc.
4. Function Type (10%): Create, analyze, transform, query, etc.

Recommendation Output Format

📋 Context Understanding:
[Brief explanation of understanding of current project state]

Main Recommendation:
- `/namespace:skill-name` - Recommendation reason
  Usage example: ...

Alternative Options:
- `/namespace:alternative1` - Description
- `/namespace:alternative2` - Description

Supporting Recommendations:
- `/namespace:complementary` - Can be used together to...

📌 Next Step Suggestions:
[Follow-up suggestions based on project progress]

🔄 Maintenance and Updates

Refresh Index

When you install new skills or want to change classification strategy:

/find-my-skills:init --refresh

View Generated Catalog

After initialization, skills-catalog.md will be generated in the current directory, where you can:

• View all installed skills
• Browse by category
• Quickly find specific functions

Custom Classification

If preset strategies don't meet your needs, you can:

1. Copy templates/custom-strategy-template.yaml
2. Modify classification dimensions and keywords
3. Re-run initialization and select your custom strategy

📚 Usage Examples

Example 1: Project Planning Stage

User: "I'm starting a bioinformatics project, what should I do first?"

AI Analysis:
- Workflow Stage: Planning & design
- Professional Domain: Bioinformatics
- Context: Project initiation

Recommendations:
Main: /boundary-lock (Define project scope)
Supporting: /scientific-skills:literature-review (Literature research)
Next Step: After planning, use /find-my-skills:implementing to find implementation tools

Example 2: Data Analysis Stage

User: "I have single-cell RNA-seq data to analyze"

AI Analysis:
- Workflow Stage: Implementation
- Professional Domain: Bioinformatics (transcriptomics)
- File Type: Biological data format
- Context: Data analysis task

Recommendations:
Main: /scientific-skills:scanpy (Standard single-cell analysis tool)
Alternative: /scientific-skills:scvi-tools (Deep learning approach)
Supporting: /scientific-skills:anndata (Data structure management)
Next Step: After analysis, use /find-my-skills:visualize for result visualization

Example 3: Paper Writing Stage

User: "Need to write a machine learning paper"

AI Analysis:
- Workflow Stage: Documentation
- Professional Domain: Machine learning + Academic research
- File Type: Academic documents
- Context: Academic writing task

Recommendations:
Main: /ml-paper-writing:20-ml-paper-writing (ML paper specialized)
Supporting:
- /scientific-skills:citation-management (Citation management)
- /scientific-skills:venue-templates (Conference templates)
- /scientific-skills:scientific-visualization (Result charts)

🛠️ Technical Details

Index Data Storage

• Location: ~/.claude/skills/find-my-skills/user-index.json
• Format: JSON
• Content: Skills metadata, classification info, strategy configuration

Catalog Document Generation

• Location: ./skills-catalog.md (current directory)
• Format: Markdown
• Content: Structured skills catalog, organized by selected strategy

Compatibility

• ✅ macOS
• ✅ Linux
• ✅ Windows (WSL)
• ✅ Cross-user environment

🆘 Troubleshooting

Issue 1: Initialization Fails

Symptoms: Error when running /find-my-skills:init

Possible Causes:

• Permission issues
• SKILL.md format errors
• Directory doesn't exist

Solutions:

# Check directory permissions
chmod +x ~/.claude/skills
chmod +x ~/.agents/skills

# Force re-initialization
/find-my-skills:init --force

Issue 2: Inaccurate Recommendations

Symptoms: Recommended skills don't match needs

Possible Causes:

• Index is outdated
• Inappropriate classification strategy
• Context understanding deviation

Solutions:

1. Refresh index: /find-my-skills:init --refresh
2. Change classification strategy
3. Provide more detailed need description

Issue 3: Cannot Find a Skill

Symptoms: You know a skill is installed, but it doesn't appear in recommendations

Possible Causes:

• Index not updated
• Keywords don't match
• Skill is in non-standard directory

Solutions:

1. Refresh index
2. Check skills-catalog.md to confirm if indexed
3. Use custom classification strategy to add relevant keywords

📖 Related Resources

• Initialization Guide: /find-my-skills:init for detailed initialization process
• Classification Strategy Templates: YAML files in templates/ directory
• Skills Catalog: Generated skills-catalog.md
• Project Plan: /Users/stephen/.claude/plans/memoized-herding-lake.md

🤝 Relationship with Other Skills

• find-skills (external): Complementary relationship, local index vs online search
• Subskill skills: Inheritance relationship, providing scenario-specific optimization
• Other skills: Recommended objects, quickly located through this skill

💡 Best Practices

1. Must initialize on first use: Ensure index data exists
2. Regularly refresh index: Update promptly after installing new skills
3. Provide sufficient context: Let AI understand your project background
4. Use subskills for quick navigation: Use subskills directly when need type is clear
5. View generated catalog: skills-catalog.md can be referenced offline
6. Custom strategies: Create your own classification when presets don't meet needs

Get Started: /find-my-skills:init 🚀