janreges/ai-distiller: Instantly Transform Large Codebases into AI-Friendly Summaries

AI Distiller is a new tool designed to enhance the performance of AI coding assistants by addressing the limitations of their context windows. Typically, AI tools struggle with large codebases, generating faulty code due to insufficient context. AI Distiller distills essential information from vast projects, compressing data to just 5-20% of its original size. It focuses on public APIs and type signatures, discarding non-critical implementation details. The system processes code in milliseconds, supports over 12 languages, and offers a high degree of configurability. Moreover, it generates structured prompts for AI agents, facilitating in-depth code analysis. Key features include integration with platforms like Claude Code and Cursor, intelligent filtering, and support for various output formats. Users can customize processing to focus on specific sections or entire repositories, significantly improving AI-assisted development speed and accuracy while reducing costs associated with token usage. Feedback is appreciated as the tool evolves.

Source link