Smart Docs MCP Server

A production-ready Model Context Protocol (MCP) server that intelligently analyzes codebases and generates comprehensive documentation. Built with TypeScript and tree-sitter for accurate code parsing.

Features

• Multi-Language Support: Analyze TypeScript, JavaScript, and Python codebases

• Smart Analysis: Extract functions, classes, methods, interfaces, types, and variables

• Documentation Coverage: Calculate documentation coverage metrics

• Missing Doc Detection: Identify undocumented code with severity levels (critical, medium, low)

• AI-Powered Suggestions: Generate documentation templates and improvement recommendations

• Markdown Output: Professional documentation in markdown format

Installation

Usage

As an MCP Server

Add to your MCP client configuration (Claude Desktop):

Windows: %APPDATA%\Claude\claude_desktop_config.json

Then restart Claude Desktop.

Available Tools

1. analyze_codebase

Analyze a codebase or file to extract code elements and calculate documentation coverage.

Input:

Output:

• Total files analyzed

• Total code elements found

• Documentation coverage percentage

• Detailed breakdown by file and element

2. generate_documentation

Generate comprehensive markdown documentation for a codebase.

Input:

Output:

• Complete markdown documentation

• Summary statistics

• Organized by file and element type

3. detect_missing_docs

Detect code elements missing documentation with severity classification.

Input:

Severity Levels:

• Critical: Public classes, interfaces, and exported functions

• Medium: Type aliases and exported types

• Low: Private methods and internal variables

Output:

• List of missing documentation by severity

• Summary statistics by severity and type

• Detailed element information

4. suggest_improvements

Analyze existing documentation and suggest improvements.

Input:

Output:

• Documentation templates for missing docs

• Suggestions for incomplete documentation

• Parameter and return value documentation hints

Project Structure

How It Works

1. Parsing: Uses tree-sitter to parse source code into Abstract Syntax Trees (AST)

2. Extraction: Identifies code elements (functions, classes, etc.) from the AST

3. Documentation Detection: Checks for JSDoc, docstrings, and inline comments

4. Analysis: Calculates coverage and detects missing documentation

5. Generation: Creates markdown documentation and improvement suggestions

Severity Classification

The server uses intelligent severity classification:

• Public APIs (classes, interfaces, exported functions) → Critical

• Type definitions and complex types → Medium

• Private methods and internal variables → Low

Error Handling

All tools include comprehensive error handling:

• Invalid paths return descriptive error messages

• Unsupported file types are gracefully skipped

• Parse errors include file location and context

Development

Requirements

• Node.js >= 18.0.0

• TypeScript 5.3+

Dependencies

• @modelcontextprotocol/sdk: MCP protocol implementation

• tree-sitter: Code parsing engine

• tree-sitter-typescript: TypeScript/JavaScript grammar

• tree-sitter-python: Python grammar

• zod: Schema validation

Testing

A test project is included in the test-project/ directory with sample files demonstrating various documentation scenarios.

License

MIT

Contributing

Contributions are welcome! Please ensure:

• Code follows TypeScript best practices

• All new features include error handling

• Documentation is updated according