Skip to content

ANALYZER.md

Latest commit

 

History

History
164 lines (120 loc) · 3.89 KB

ANALYZER.md

File metadata and controls

164 lines (120 loc) · 3.89 KB

Skill Analyzer Setup

This repository includes the skill-seekers codebase analyzer for generating comprehensive AI skills from any codebase.

Quick Start

1. Install the Analyzer (One-Time)

make install

This creates a Python virtual environment and installs skill-seekers.

2. Generate Skills

For Avni Repositories

# Analyze avni-client
make analyze-avni-client

# Analyze avni-models
make analyze-avni-models

For Any Repository

# Basic usage
make analyze REPO=/path/to/repository

# Custom output location
make analyze REPO=/path/to/repository OUTPUT=skills/my-skill-name

What Gets Generated

Each analysis creates a comprehensive skill with:

Core Documentation

  • SKILL.md - Main skill overview with statistics and quick reference
  • code_analysis.json - Raw analysis data

API Reference (178-564 files)

  • Complete API documentation for all classes, functions, and modules
  • Method signatures, parameters, return types
  • Extracted from source code comments and structure

Dependency Analysis

  • dependency_graph.json - Full dependency relationships
  • dependency_graph.dot - GraphViz format
  • dependency_graph.mmd - Mermaid diagram format
  • statistics.json - Dependency metrics

Design Patterns

  • detected_patterns.json - Factory, Singleton, Observer, Strategy, etc.
  • Confidence scores for each pattern
  • File locations and evidence

Test Examples

  • test_examples.json - Extracted test code examples
  • test_examples.md - Formatted examples with complexity scores
  • Real-world usage patterns

Configuration Patterns

  • config_patterns.json - Configuration analysis
  • config_patterns.md - Formatted config documentation
  • Settings extraction from JSON, YAML, TOML, ENV files

Architecture Analysis

  • architectural_patterns.json - MVC, MVVM, Layered, etc.
  • Component mapping (Models, Views, Controllers)
  • Confidence scores and evidence

Project Documentation

  • documentation_index.json - Catalog of all docs
  • Extracted markdown files organized by category
  • README, guides, and other documentation

Example: Analyzing a New Repository

# Clone or locate the repository
cd /path/to/repositories

# Generate skill
cd /path/to/agent-skills
make analyze REPO=/path/to/repositories/my-project OUTPUT=skills/my-project

# Commit the generated skill
git add skills/my-project/
git commit -m "Add my-project skill"
git push

Analysis Depth

The Makefile uses --depth full which includes:

  • ✅ API Reference extraction
  • ✅ Dependency graph generation
  • ✅ Design pattern detection
  • ✅ Test example extraction
  • ✅ Configuration pattern analysis
  • ✅ Architectural pattern detection
  • ✅ Project documentation extraction

Output Size

Typical skill sizes:

  • avni-client: ~16MB (566 files, 564 API refs)
  • avni-models: ~2MB (178 files, 178 API refs)

Cleanup

Remove the virtual environment:

make clean

Advanced Usage

For custom analysis options, use the analyzer directly:

source venv/bin/activate
skill-seekers-codebase --help

Available options:

  • --depth {surface,deep,full} - Analysis depth
  • --languages - Filter by languages
  • --file-patterns - Filter by file patterns
  • --skip-* - Skip specific analysis steps
  • --verbose - Detailed logging

Troubleshooting

Virtual environment not found

make install

Analysis fails

Check that the repository path is correct and accessible:

ls -la /path/to/repository

Python version issues

Requires Python 3.8+:

python3 --version

Integration with AI Tools

Generated skills can be used with:

  • Claude Code - Via skills.sh integration
  • Windsurf - Local skill loading
  • Custom AI tools - JSON/Markdown format

The comprehensive documentation enables AI assistants to understand your codebase deeply and provide accurate, context-aware assistance.