Valkey Vector Database

You can use Valkey as a vector database with Agno using the valkey-search module.

Setup

Valkey vector search requires the valkey-search module. Use the valkey/valkey-bundle Docker image which includes it:

docker run -d --name my-valkey \
  -p 6379:6379 \
  valkey/valkey-bundle

Example

agent_with_knowledge.py

from agno.agent import Agent
from agno.knowledge.knowledge import Knowledge
from agno.vectordb.valkey import ValkeyDB
from agno.vectordb.search import SearchType

# Initialize Valkey Vector DB
vector_db = ValkeyDB(
    index_name="agno_docs",
    host="localhost",
    port=6379,
    search_type=SearchType.vector,
)

# Build a Knowledge base backed by Valkey
knowledge = Knowledge(
    name="My Valkey Knowledge Base",
    description="Knowledge base using Valkey as the vector store",
    vector_db=vector_db,
)

# Add content
knowledge.insert(
    name="Recipes",
    url="https://agno-public.s3.amazonaws.com/recipes/ThaiRecipes.pdf",
    metadata={"category": "recipe_book"},
    skip_if_exists=True,
)

# Query with an Agent
agent = Agent(knowledge=knowledge)
agent.print_response("List down the ingredients to make Massaman Gai", markdown=True)

Valkey Params

Parameter	Type	Default	Description
`index_name`	`str`	Required	Name of the Valkey search index to store vector data
`host`	`str`	`"localhost"`	Valkey server host
`port`	`int`	`6379`	Valkey server port
`username`	`Optional[str]`	`None`	Username for authentication
`password`	`Optional[str]`	`None`	Password for authentication
`use_tls`	`bool`	`False`	Enable TLS encryption
`database_id`	`Optional[int]`	`None`	Logical database index (e.g. 0-15)
`request_timeout`	`Optional[int]`	`None`	Milliseconds to wait for a request to complete. If unset, the GLIDE client default (250 ms) applies
`client_name`	`str`	`"agno_vectordb_client"`	Connection name, visible in `CLIENT LIST`
`glide_client`	`Optional[GlideClient]`	`None`	Pre-configured Valkey GLIDE client instance
`embedder`	`Optional[Embedder]`	`None`	Embedder instance to generate embeddings (defaults to `OpenAIEmbedder()` when unset)
`search_type`	`SearchType`	`SearchType.vector`	Type of search to perform (vector, keyword)
`distance`	`Distance`	`Distance.cosine`	Distance metric (cosine, l2, max_inner_product)
`vector_algorithm`	`str`	`"HNSW"`	Vector index algorithm (HNSW or FLAT)
`reranker`	`Optional[Reranker]`	`None`	Reranker for search results
`id`	`Optional[str]`	`None`	Optional custom ID. If not provided, an ID will be generated
`name`	`Optional[str]`	`None`	Optional name for the vector database
`description`	`Optional[str]`	`None`	Optional description for the vector database

​Setup

​Example

​Valkey Params

Setup

Example

Valkey Params