Skip to main content

clawdesk-rag

The RAG crate implements Retrieval-Augmented Generation — a technique that gives AI agents access to your documents and knowledge base when answering questions.

What It Does (Plain English)​

Imagine you have a company handbook, product documentation, or personal notes. Without RAG, the AI doesn't know about any of this. With RAG, when you ask a question, ClawDesk first searches your documents for relevant information, then gives that context to the AI along with your question. The result: accurate answers based on YOUR data.

Key Features​

  • Document ingestion — Process PDFs, text files, markdown, and more
  • Chunking — Splits documents into searchable pieces
  • Vector indexing — Enables semantic search (search by meaning, not just keywords)
  • Hybrid search — Combines keyword (BM25) and vector search for best results
  • Context assembly — Selects the most relevant chunks to include with your question

How It Works​

Architecture Role​

LayerPosition
ServicesKnowledge retrieval and augmentation

Dependencies​

  • clawdesk-types — Document and chunk types
  • clawdesk-memory — Hybrid search infrastructure
  • clawdesk-sochdb — Vector storage