Third Party Tools

{ title="This feature is currently available for Python. Java support is planned/ coming soon."}

ADK is designed to be highly extensible, allowing you to seamlessly integrate tools from other AI Agent frameworks like CrewAI and LangChain. This interoperability is crucial because it allows for faster development time and allows you to reuse existing tools.

1. Using LangChain Tools

ADK provides the LangchainTool wrapper to integrate tools from the LangChain ecosystem into your agents.

Example: Web Search using LangChain's Tavily tool

Tavily provides a search API that returns answers derived from real-time search results, intended for use by applications like AI agents.

Follow ADK installation and setup guide.
Install Dependencies: Ensure you have the necessary LangChain packages installed. For example, to use the Tavily search tool, install its specific dependencies:
```
pip install langchain_community tavily-python
```
Obtain a Tavily API KEY and export it as an environment variable.
```
export TAVILY_API_KEY=<REPLACE_WITH_API_KEY>
```

Import: Import the LangchainTool wrapper from ADK and the specific LangChain tool you wish to use (e.g, TavilySearchResults).

from google.adk.tools.langchain_tool import LangchainTool
from langchain_community.tools import TavilySearchResults

Instantiate & Wrap: Create an instance of your LangChain tool and pass it to the LangchainTool constructor.

# Instantiate the LangChain tool
tavily_tool_instance = TavilySearchResults(
    max_results=5,
    search_depth="advanced",
    include_answer=True,
    include_raw_content=True,
    include_images=True,
)

# Wrap it with LangchainTool for ADK
adk_tavily_tool = LangchainTool(tool=tavily_tool_instance)

Add to Agent: Include the wrapped LangchainTool instance in your agent's tools list during definition.

from google.adk import Agent

# Define the ADK agent, including the wrapped tool
my_agent = Agent(
    name="langchain_tool_agent",
    model="gemini-2.0-flash",
    description="Agent to answer questions using TavilySearch.",
    instruction="I can answer your questions by searching the internet. Just ask me anything!",
    tools=[adk_tavily_tool] # Add the wrapped tool here
)

Full Example: Tavily Search

Here's the full code combining the steps above to create and run an agent using the LangChain Tavily search tool.

--8<-- "examples/python/snippets/tools/third-party/langchain_tavily_search.py"

2. Using CrewAI tools

ADK provides the CrewaiTool wrapper to integrate tools from the CrewAI library.

Example: Web Search using CrewAI's Serper API

Serper API provides access to Google Search results programmatically. It allows applications, like AI agents, to perform real-time Google searches (including news, images, etc.) and get structured data back without needing to scrape web pages directly.

Follow ADK installation and setup guide.
Install Dependencies: Install the necessary CrewAI tools package. For example, to use the SerperDevTool:
```
pip install crewai-tools
```
Obtain a Serper API KEY and export it as an environment variable.
```
export SERPER_API_KEY=<REPLACE_WITH_API_KEY>
```

Import: Import CrewaiTool from ADK and the desired CrewAI tool (e.g, SerperDevTool).

from google.adk.tools.crewai_tool import CrewaiTool
from crewai_tools import SerperDevTool

Instantiate & Wrap: Create an instance of the CrewAI tool. Pass it to the CrewaiTool constructor. Crucially, you must provide a name and description to the ADK wrapper, as these are used by ADK's underlying model to understand when to use the tool.

# Instantiate the CrewAI tool
serper_tool_instance = SerperDevTool(
    n_results=10,
    save_file=False,
    search_type="news",
)

# Wrap it with CrewaiTool for ADK, providing name and description
adk_serper_tool = CrewaiTool(
    name="InternetNewsSearch",
    description="Searches the internet specifically for recent news articles using Serper.",
    tool=serper_tool_instance
)

Add to Agent: Include the wrapped CrewaiTool instance in your agent's tools list.

from google.adk import Agent

# Define the ADK agent
my_agent = Agent(
    name="crewai_search_agent",
    model="gemini-2.0-flash",
    description="Agent to find recent news using the Serper search tool.",
    instruction="I can find the latest news for you. What topic are you interested in?",
    tools=[adk_serper_tool] # Add the wrapped tool here
)

Full Example: Serper API

Here's the full code combining the steps above to create and run an agent using the CrewAI Serper API search tool.

--8<-- "examples/python/snippets/tools/third-party/crewai_serper_search.py"

5.2 KiB Raw Permalink Blame History