Visual Agent Builder · DataStax

Add MCP Tools to Langflow Visual Agent Flows

Langflow's MCP component lets you connect any MCP server to your visual agent pipelines. Wire Agent MCP Studio's relay bridge into Langflow to call browser-built Python, SQL, and API tools from any flow without running a local server.

Share:

Prerequisites

Step-by-Step Setup

  1. Start bridge.js on the Langflow server

    curl -O https://agentmcp.studio/bridge.js && npm install ws
    node bridge.js wss://agentmcp.studio/api/relay/YOUR-UUID

    Open the studio → SettingsMCP Relay Bridge → click Connect.

  2. Add an MCP Tools component to your flow

    In Langflow's flow editor, search for MCP Tools in the component sidebar. Drag it into your canvas. Configure:

    • Server URL: Leave as stdio
    • Command: node
    • Args: /path/to/bridge.js wss://agentmcp.studio/api/relay/YOUR-UUID

    The component will list all available tools from Agent MCP Studio once connected.

  3. Connect to an Agent component

    Wire the MCP Tools component's output to an Agent or Tool Calling Agent component's Tools input. Connect a Chat Input and Chat Output to complete the flow.

  4. Run the flow

    Click Run or use the Playground to test. Ask the agent to use one of your browser tools — Langflow routes the tool call through bridge.js to your browser tab.

Frequently Asked Questions

Yes — Langflow (the visual LangChain builder) includes an MCP Tools component. Add it to your flow, configure the server connection details, and your Agent MCP Studio tools become usable nodes in any Langflow agent flow.

Langflow is a visual, drag-and-drop builder for LangChain applications. Instead of writing Python, you connect components visually. It uses LangChain under the hood, so LangChain MCP support is accessible via Langflow's MCP component.

Yes — Langflow supports Ollama as an LLM provider. Combine the Ollama component with the MCP Tools component and an Agent component to build a fully local agent that calls your browser tools.

For Langflow Cloud, bridge.js needs to run on a machine accessible to the Langflow Cloud backend. The MCP connection uses stdio (subprocess), so bridge.js must be on the Langflow server or accessible via SSH/API.

Use Langflow's Playground to test flows step by step. Enable debug mode to see each tool call and response. You can also monitor the bridge.js process output to see incoming MCP messages from Langflow.

Related Integrations