Swarmauri Query Knowledge Base Tool

swarmauri_tool_queryknowledgebase exposes ranked knowledge-base retrieval as a Swarmauri tool while preserving document content, metadata, identifiers, and scores.

Features

• Uses the repository-standard retrieve(query, top_k) workflow.
• Returns structured, ranked results instead of flattening documents into text.
• Supports exact-match metadata filtering and bounded result counts.
• Keeps runtime knowledge-base adapters out of serialized tool state.
• Wraps provider failures without leaking connection details.

Installation

uv add swarmauri_tool_queryknowledgebase

pip install swarmauri_tool_queryknowledgebase

Usage

from swarmauri_tool_queryknowledgebase import QueryKnowledgeBaseTool

tool = QueryKnowledgeBaseTool(knowledge_base=my_document_store)
results = tool(
    "How are API keys rotated?",
    top_k=3,
    metadata_filter={"product": "sdk"},
)

for result in results:
    print(result["rank"], result["content"], result["metadata"])

The configured knowledge base must implement retrieve(query, top_k=5) and return documents or dictionaries in relevance order.

Project Links

• Swarmauri SDK
• Contributing guide

License

Apache-2.0.