Implementation:CrewAIInc CrewAI LlamaIndex Tool

Knowledge Sources	CrewAI
Domains	Tools, LlamaIndex, Integration
Last Updated	2026-02-11 00:00 GMT

Overview

LlamaIndexTool wraps LlamaIndex tools and query engines to make them compatible with CrewAI agents.

Description

The LlamaIndexTool extends BaseTool and serves as a critical bridge between the CrewAI and LlamaIndex ecosystems. It stores a reference to a LlamaIndex tool instance via the llama_index_tool attribute. The _run method executes the wrapped tool, optionally extracting just the content from responses when result_as_answer is enabled. Two class factory methods provide flexible construction: from_tool() validates and wraps a LlamaIndex BaseTool instance by extracting its metadata (name, description) and fn_schema for argument validation; from_query_engine() wraps a BaseQueryEngine with a custom QueryToolSchema that remaps the default input parameter to query (to avoid confusion within CrewAI), and sets resolve_input_errors=True to handle the schema mismatch gracefully.

Usage

Use this tool when you need to leverage LlamaIndex's extensive ecosystem of tools, query engines, and data connectors within a CrewAI agent workflow.

Code Reference

Source Location

• Repository: CrewAI
• File: lib/crewai-tools/src/crewai_tools/tools/llamaindex_tool/llamaindex_tool.py
• Lines: 1-92

Signature

class LlamaIndexTool(BaseTool):
    llama_index_tool: Any

    def _run(self, *args: Any, **kwargs: Any) -> Any: ...

    @classmethod
    def from_tool(cls, tool: Any, **kwargs: Any) -> LlamaIndexTool: ...

    @classmethod
    def from_query_engine(
        cls,
        query_engine: Any,
        name: str | None = None,
        description: str | None = None,
        return_direct: bool = False,
        **kwargs: Any,
    ) -> LlamaIndexTool: ...

Import

from crewai_tools import LlamaIndexTool

I/O Contract

Inputs

Name	Type	Required	Description
llama_index_tool	Any	Yes	A LlamaIndex BaseTool instance to wrap
query_engine	Any	Yes (for from_query_engine)	A LlamaIndex BaseQueryEngine instance
name	str or None	No	Tool name override (for from_query_engine)
description	str or None	No	Tool description override (for from_query_engine)
return_direct	bool	No	Whether to return the query engine result directly; defaults to False

Outputs

Name	Type	Description
_run() returns	Any	The result from the wrapped LlamaIndex tool; returns content string if result_as_answer is True

Usage Examples

From LlamaIndex Tool

from crewai_tools import LlamaIndexTool
from llama_index.core.tools import FunctionTool

llama_tool = FunctionTool.from_defaults(fn=my_function)
tool = LlamaIndexTool.from_tool(llama_tool)

From Query Engine

from crewai_tools import LlamaIndexTool
from llama_index.core import VectorStoreIndex

index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine()
tool = LlamaIndexTool.from_query_engine(
    query_engine,
    name="Document Search",
    description="Search through indexed documents",
)

Related Pages

• Principle:CrewAIInc_CrewAI_Built_In_Tool_Selection