Skip to content

AgentVault LangChain Integration Example

This directory contains a basic example demonstrating how to integrate an AgentVault A2A compliant agent as a tool within the LangChain framework.

Components

  • requirements.txt: Defines the necessary Python dependencies (langchain-core, agentvault, httpx).
  • a2a_tool.py: Contains the A2AAgentTool class, which inherits from LangChain's BaseTool. This tool handles the interaction with a remote A2A agent using the agentvault client library.
  • example_usage.py: A simple script showing how to instantiate and invoke the A2AAgentTool.

Setup

  1. Install Dependencies: Navigate to this directory (examples/langchain_integration) in your terminal and install the requirements. This includes installing the local agentvault library in editable mode. 

    pip install -r requirements.txt
    Note: This assumes your virtual environment is activated and you are in the correct directory. The -e ../../agentvault_library line installs the library from your local source tree.

  2. Configure Agent Reference: Open example_usage.py and modify the EXAMPLE_AGENT_REF variable to point to a valid agent:

    • Agent ID: If using an ID from a running AgentVault Registry (e.g., test-org/my-agent), ensure the AGENTVAULT_REGISTRY_URL environment variable is set correctly or modify the default in a2a_tool.py.
    • Agent URL: Provide the direct URL to the agent's agent-card.json (e.g., http://localhost:8001/agent-card.json).
    • Local File: Provide the path to a local agent-card.json file (e.g., ./path/to/your/agent-card.json).

  3. Configure Credentials: If the target agent requires authentication (e.g., apiKey or oauth2), ensure the necessary credentials are set up using the AgentVault CLI (agentvault config set ...) or environment variables, matching the service_identifier specified in the agent's card.

Running the Example

Once set up, you can run the example script:

python example_usage.py

The script will:

  1. Instantiate the A2AAgentTool.
  2. Prepare the input (agent reference and text prompt).
  3. Invoke the tool's _arun method.
  4. The tool will internally use AgentVaultClient to:
    • Load the agent card.
    • Initiate a task.
    • Stream events (though only assistant text messages are captured in this basic example).
    • Wait for the task to complete.
  5. Print the final aggregated text response from the agent.

Note: This example primarily demonstrates the tool's structure and integration. For it to fully succeed, you need a running A2A agent accessible at the specified EXAMPLE_AGENT_REF that can handle the input prompt. You can use the agentvault-server-sdk examples or your own agent implementation.