ai-distiller

Category: repository-code-analysis-mcp-servers
Tags: code-analysis, ai-assistant, documentation
Source: https://github.com/janreges/ai-distiller
License: MIT

Description

ai-distiller is an open-source tool that analyzes large codebases and extracts only the essential public APIs, types, and structural information into a compact, AI-digestible format. It is commonly used as an MCP-accessible utility so AI agents can more effectively understand and work with complex repositories.

Features

• Extracts essential public APIs, types, and structural information from large codebases
• Produces highly compressed, AI-friendly context (approximately 90–98% code size reduction)
• Designed for fast processing on large repositories
• Open-source, licensed under MIT
• Usable as a command-line (CLI) tool
• MCP integration for use as an MCP server in AI workflows
• Intended for AI assistants/agents to better navigate and reason about code
• Supports 12+ programming languages (multi-language code analysis)
• Suitable for generating distilled documentation-like views of a project’s structure
• Can be integrated into AI prompt workflows to reduce context size and cost while maintaining essential information

Integrations & Interfaces

• CLI: Direct command-line usage for local or CI workflows
• MCP (Model Context Protocol): Accessible as an MCP server, enabling AI tools/agents to query distilled repository structure

Use Cases

• Preparing large repositories for efficient AI-assisted code understanding and navigation
• Reducing context size and token costs when sending code information to LLMs
• Generating concise representations of public APIs and types for documentation or review

Pricing

ai-distiller is open-source under the MIT license and can be used at no monetary cost. There are no stated paid plans in the provided content.