Implementation:CrewAIInc CrewAI Weaviate Vector Search Tool

Knowledge Sources	CrewAI
Domains	Tools, Vector_Database, RAG
Last Updated	2026-02-11 00:00 GMT

Overview

WeaviateVectorSearchTool performs hybrid vector searches against a Weaviate cloud vector database to retrieve semantically relevant documents.

Description

The WeaviateVectorSearchTool extends BaseTool and requires collection_name, weaviate_cluster_url, and weaviate_api_key at initialization. It uses default factory functions for the vectorizer (text2vec_openai with nomic-embed-text model) and generative model (openai gpt-4o), both of which can be overridden. On initialization, it retrieves the OPENAI_API_KEY from the environment for the Weaviate client headers. If the weaviate-client package is missing, it prompts for installation via click.confirm. The _run method connects to Weaviate cloud, retrieves the specified collection (or creates it if not found), performs a hybrid query with configurable limit (default 3) and alpha (default 0.75, balancing keyword and vector search), serializes results to JSON, then closes the client connection.

Usage

Use this tool when a CrewAI agent needs to query an enterprise-grade Weaviate vector database for semantically relevant internal documents using hybrid search combining keyword and vector similarity.

Code Reference

Source Location

• Repository: CrewAI
• File: lib/crewai-tools/src/crewai_tools/tools/weaviate_tool/vector_search.py
• Lines: 1-138

Signature

class WeaviateToolSchema(BaseModel):
    query: str = Field(..., description="The query to search retrieve relevant information ...")

class WeaviateVectorSearchTool(BaseTool):
    name: str = "WeaviateVectorSearchTool"
    description: str = "A tool to search the Weaviate database for relevant information on internal documents."
    args_schema: type[BaseModel] = WeaviateToolSchema
    collection_name: str = Field(description="The name of the Weaviate collection to search")
    limit: int | None = Field(default=3)
    alpha: float = Field(default=0.75)
    weaviate_cluster_url: str = Field(...)
    weaviate_api_key: str = Field(...)
    vectorizer: Any = Field(default_factory=_set_vectorizer)
    generative_model: Any = Field(default_factory=_set_generative_model)
    env_vars: list[EnvVar]  # OPENAI_API_KEY

    def _run(self, query: str) -> str:
        ...

Import

from crewai_tools import WeaviateVectorSearchTool

I/O Contract

Inputs

Name	Type	Required	Description
query	str	Yes	The search query to retrieve relevant information from Weaviate

Outputs

Name	Type	Description
_run() returns	str	JSON string containing the properties of matching objects from the Weaviate collection

Usage Examples

Basic Usage

from crewai_tools import WeaviateVectorSearchTool

tool = WeaviateVectorSearchTool(
    collection_name="InternalDocs",
    weaviate_cluster_url="https://my-cluster.weaviate.network",
    weaviate_api_key="my-weaviate-api-key",
)
result = tool._run(query="quarterly revenue report")

Related Pages

• Principle:CrewAIInc_CrewAI_Built_In_Tool_Selection