Show HN: Llmdoc – annotate codebase with LLM summaries only re-scan what changed (github.com)

Llmdoc, a new tool introduced on HN, revolutionizes how developers annotate their codebases by leveraging large language models (LLMs) to generate concise summaries. It scans the codebase for changes using SHA-256 hashes, ensuring that only modified files request new LLM summaries, thus optimizing token usage and reducing costs. This is particularly significant for the AI/ML community as it streamlines the integration of AI tools into existing workflows while maintaining the legibility and accessibility of the codebase. Llmdoc offers features like two storage modes—inline, where summaries are stored as comments in source files, and index, where they are saved in a separate YAML file—ensuring that annotations are either co-located for visibility or kept pristine. With support for over 50 programming languages and compatibility with major LLM providers like Anthropic and OpenAI, Llmdoc is easily integrated into CI/CD pipelines to enforce the accuracy of code annotations. This capability not only improves code readability for teams and AI models but also expedites the software development process by keeping documentation up-to-date without unnecessary API calls.