01Offload large file queries to a local LLM (Ollama) for context compression
020 GitHub stars
03Achieve up to 30x token savings for large file interactions
04Utilize `local_answer` for efficient Q&A on files without loading them into Claude's context
05Compress large in-context text blocks with `local_summarize` for reuse
06Extract specific, cited passages from documents using `local_extract`