code-analyzer

name	code-analyzer
description	Analyze code structure - functions, classes, imports, call relationships, community detection. Pure code analysis, no LLM needed.
trigger	/code-analyzer

/code-analyzer

Analyze a codebase and extract structural information: files, functions, classes, imports, call relationships, and community clusters.

Usage

/code-analyzer <path>                    # Full analysis → JSON output
/code-analyzer <path> -o json            # Output JSON to stdout
/code-analyzer <path> -o file           # Write to code_graph.json
/code-analyzer <path> -o both           # Both stdout and file (default)
/code-analyzer <path> -t calls          # Only call relationships
/code-analyzer <path> -t definitions    # Only function/class definitions
/code-analyzer <path> -t all            # All output types (default)
/code-analyzer <path> -u                # Incremental update (only changed files)
/code-analyzer <path> --langs python,js  # Filter by languages
/code-analyzer <path> -v                # Verbose output

What code-analyzer provides

• structure: Project file tree with line counts
• definitions: All functions and classes with file:line locations
• calls: Function call relationships (who calls whom)
• imports: Import/export relationships between files
• clusters: Community detection grouping related files
• stats: Statistics (files, functions, classes, lines of code)

When to use

• Understanding a new codebase quickly
• Finding which functions call which other functions
• Identifying the main classes and their relationships
• Finding circular dependencies or tightly coupled modules
• When you need code structure without reading every file

Output format

JSON with the following structure:

{
  "meta": {
    "path": "/path/to/project",
    "timestamp": "1234567890",
    "languages": ["python", "javascript"],
    "incremental": false
  },
  "structure": {
    "files": [{"path": "src/main.py", "size": 1234, "lines": 100, "language": "python"}],
    "directories": ["src", "tests"]
  },
  "definitions": {
    "functions": [{"name": "main", "file": "src/main.py", "line": 10, "params": ["arg1"]}],
    "classes": [{"name": "Config", "file": "src/config.py", "line": 5, "methods": ["load", "save"]}]
  },
  "calls": {
    "edges": [{"from": "src/main.py:main", "to": "src/utils.py:helper", "type": "direct"}]
  },
  "imports": {
    "edges": [{"from": "src/main.py", "to": "src/config.py", "symbol": "Config"}]
  },
  "clusters": [
    {"id": 0, "members": ["src/main.py", "src/utils.py"], "cohesion": 0.85}
  ],
  "stats": {
    "total_files": 50,
    "total_functions": 120,
    "total_classes": 30,
    "lines_of_code": 5000
  }
}

Supported languages

Python, JavaScript, TypeScript, Java, C, C++, Go, Rust

Examples

Quick overview

/code-analyzer ./myproject

Get only call relationships

/code-analyzer ./myproject -t calls -o json

Find all classes and methods

/code-analyzer ./myproject -t definitions

Incremental update (after code changes)

/code-analyzer ./myproject -u

Notes

• This tool does pure code analysis using regex/parsing - no LLM involved
• It's fast and suitable for large codebases
• For semantic understanding (why this code exists, design rationale), use the AI Agent directly
• The JSON output can be passed to an AI Agent for further analysis or to build a knowledge graph