35 lines
1.2 KiB
Python
35 lines
1.2 KiB
Python
"""Tools for the RAG agent.
|
||
|
||
Two tools are exposed via the ``@tool`` decorator:
|
||
* ``search_knowledge_base`` – semantic search in the Qdrant vector store.
|
||
* ``add_to_knowledge_base`` – add a document to the vector store.
|
||
"""
|
||
|
||
from typing import List, Dict
|
||
|
||
from langchain.tools import tool
|
||
|
||
from .vector_store import add_document, search_text
|
||
|
||
# ---------------------------------------------------------------------------
|
||
# Tool definitions
|
||
# ---------------------------------------------------------------------------
|
||
|
||
@tool("search_knowledge_base")
|
||
async def search_knowledge_base(query: str, max_results: int = 5) -> List[str]:
|
||
"""Return the top *max_results* relevant chunks for *query*.
|
||
|
||
The function is asynchronous because LangChain expects async tools when
|
||
the agent runs in an async context. The underlying vector store calls
|
||
are synchronous, so we simply wrap the result.
|
||
"""
|
||
return search_text(query, k=max_results)
|
||
|
||
@tool("add_to_knowledge_base")
|
||
async def add_to_knowledge_base(content: str, title: str) -> str:
|
||
"""Add *content* under *title* to the knowledge base.
|
||
|
||
Returns a confirmation string.
|
||
"""
|
||
add_document(title, content)
|
||
return f"Document '{title}' added to knowledge base." |