janreges/ai-distiller-mcp

Overview

An ultra-fast, open-source tool written in C that extracts only essential public APIs, types, and structure from large codebases. It compresses code by 90–98% to produce AI-friendly context, integrating via CLI or MCP (Model Context Protocol) and supporting 12+ languages.

Best for

Best for
Developers who need to feed large codebases into LLMs cheaply and efficiently

Use cases

• Reduce token usage when sending large codebases to LLMs
• Extract clean public interfaces for AI-assisted code generation or analysis
• Prepare minimal code context for cost-effective prompt workflows

How to use

Install

npm install -g @janreges/ai-distiller-mcp

Tools exposed

• prompt-for-refactoring-suggestion
• prompt-for-complex-codebase-analysis
• prompt-for-security-analysis
• prompt-for-performance-analysis
• prompt-for-best-practices-analysis
• prompt-for-bug-hunting
• prompt-for-single-file-docs
• prompt-for-diagrams
• flow-for-deep-file-to-file-analysis
• flow-for-multi-file-docs
• visual-progress-bar
• stock-ticker
• speedometer-dashboard
• minimalist-sparkline
• ci-friendly

Tested with

Claude Desktop, Claude Code, Cursor, Windsurf, VS Code, ChatGPT

Notes

An ultra-fast, open-source tool written in C that extracts only essential public APIs, types, and structure from large codebases. It compresses code by 90–98% to produce AI-friendly context, integrating via CLI or MCP (Model Context Protocol) and supporting 12+ languages.

159 stars on GitHub. Last updated 2026-05-25. Licensed MIT.

Use cases

• Reduce token usage when sending large codebases to LLMs
• Extract clean public interfaces for AI-assisted code generation or analysis
• Prepare minimal code context for cost-effective prompt workflows

Pros

• Extremely fast due to implementation in C
• Open source with no vendor lock-in
• Significant token savings (90–98% compression)

Cons

• Only captures public APIs and types, losing internal implementation details
• Limited community adoption (159 stars) may mean fewer integrations or fewer tested edge cases
• Requires familiarity with MCP or CLI for integration

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.