LanceDB | Agno

LanceDb supports vector, full-text, and hybrid search. A filesystem path creates a local database. A db:// URI connects to LanceDB Cloud and requires api_key= or LANCEDB_API_KEY.

In the current adapter, dictionary filters are applied after LanceDB returns the limited candidates, and filter-expression lists are ignored. Filtered searches can return fewer documents than requested. Search results also omit the stored LanceDB row ID.

from agno.agent import Agent
from agno.knowledge.knowledge import Knowledge
from agno.vectordb.lancedb import LanceDb

knowledge = Knowledge(
    vector_db=LanceDb(
        table_name="recipes",
        uri="tmp/lancedb",
    )
)
agent = Agent(knowledge=knowledge, search_knowledge=True)

if __name__ == "__main__":
    knowledge.insert(
        url="https://agno-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf"
    )
    agent.print_response("How do I make pad thai?", markdown=True)

Run the Example

Set up your virtual environment

uv venv --python 3.12
source .venv/bin/activate

Install dependencies

uv pip install -U agno lancedb openai pypdf

Export the API key

export OPENAI_API_KEY=your_openai_api_key_here

Run the example

python agent_with_knowledge.py

Search Modes

`search_type`	Behavior
`SearchType.vector`	Searches by embedding similarity
`SearchType.keyword`	Uses LanceDB native full-text search
`SearchType.hybrid`	Runs LanceDB hybrid vector and full-text search

Parameters

Parameter	Type	Default	Description
`uri`	`str`	`/tmp/lancedb`	The URI to connect to.
`table`	`LanceTable`	-	The Lance table to use.
`table_name`	`str`	-	The name of the table to use.
`connection`	`DBConnection`	-	The database connection to use.
`api_key`	`str`	-	The API key to use.
`embedder`	`Embedder`	-	The embedder to use.
`search_type`	`SearchType`	vector	The search type to use.
`distance`	`Distance`	cosine	The distance to use.
`nprobes`	`int`	-	The number of probes to use. More Info
`reranker`	`Reranker`	-	The reranker to use. More Info
`on_bad_vectors`	`str`	-	What to do if the vector is bad. One of `"error"`, `"drop"`, `"fill"`, `"null"`.
`fill_value`	`float`	-	The value to fill the vector with if `on_bad_vectors` is `"fill"`.
`async_connection`	`Optional[lancedb.AsyncConnection]`	`None`	Existing async connection to use.
`async_table`	`Optional[lancedb.db.AsyncTable]`	`None`	Existing async table to use.
`name`	`Optional[str]`	`None`	Name of the vector database.
`description`	`Optional[str]`	`None`	Description of the vector database.
`id`	`Optional[str]`	Generated	ID derived from the URI and table name when omitted.

Next Steps

Task	Guide
Insert, search, and delete content	LanceDB usage
Call async Agno methods	Async LanceDB usage
Configure hybrid search	LanceDB hybrid search

Developer Resources

LanceDB Python SDK