Skip to main content

KnowledgeRepo

Performs a semantic search over your knowledge repository backed by Qdrant and embeddings.

Usage​

Provide QDRANT_COLLECTION_NAME, QDRANT_URL, KR_API_KEY, and OPENAI_API_KEY in the tool config. Pass a query string to search the repository.

{ "query": "How to configure SVAHNAR agent network" }

Returns​

Top matching text snippets from the knowledge repository (by default the top 3 results).

YAML usage​

tools:
  tool_assigned:
    - name: "KnowledgeRepo"
      config:
        QDRANT_COLLECTION_NAME: "my_qdrant_collection"
        QDRANT_URL: "https://qdrant.example.com"
        KR_API_KEY: "${KR_API_KEY}"
        OPENAI_API_KEY: "${OPENAI_API_KEY}"

Invocation payload:

payload:
	query: "How to configure SVAHNAR agent network"
	k_repo_id: "<repo-uuid>" # optional to scope the search

More details​

This tool uses OpenAI embeddings to create a query vector and Qdrant to perform the nearest-neighbor search; result ordering and quality rely on the embedding model selected.

Agent integration example​

create_vertical_agent_network:
  agent-1:
    agent_name: "knowledge_searcher"
    LLM_config:
      params:
        model: "gpt-4o-mini"
    tools:
      tool_assigned:
        - name: "KnowledgeRepo"
          config:
            QDRANT_COLLECTION_NAME: "my_qdrant_collection"
            QDRANT_URL: "https://qdrant.example.com"
            KR_API_KEY: "${KR_API_KEY}"
            OPENAI_API_KEY: "${OPENAI_API_KEY}"
    agent_function:
      - "Search the knowledge repository for relevant passages and return top snippets."