kb-ingest

Ingesting new sources into a knowledge base involves resolving local files, URLs, or pasted content, classifying the material, and surgically updating affected library files while logging changes to maintain provenance and enable precise future retrieval.

Surgical updates to touched library files ensure unrelated knowledge base content remains undisrupted while integrating new sources. This approach maintains provenance and automatically rebuilds the shelf-index for precise retrieval.

Logging and provenance during a knowledge base update work by appending changes to log.md. This creates an auditable trace of ingested sources, ensuring documentation and traceability for future retrieval and updates.

Yes, you can ingest URLs and pasted text into a local file knowledge base. The ingestion process resolves these diverse source types, determines their relevance, and routes the content to appropriate library files.

Yes, you need a pre-configured knowledge base and the agent-knowledge-updater available. These prerequisites are required to classify content, apply surgical file updates, rebuild the shelf-index, and log changes.

Your knowledge base shelf-index needs rebuilding after adding new sources to maintain accurate documentation and traceability. Rebuilding ensures the index reflects the latest ingested content for precise future retrieval.