WxO Agent

A lightweight MCP server that invokes a single Watson Orchestrate agent via environment or config.

javascript

0

GitHub Stars

javascript

Language

3 months ago

First Indexed

2 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "markusvankempen-wxo-agent-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "wxo-agent-mcp"
      ],
      "env": {
        "WO_API_KEY": "your-api-key",
        "WO_AGENT_ID": "your-agent-id",
        "WO_INSTANCE_URL": "https://your-instance-id.orchestrate.ibm.com"
      }
    }
  }
}

WxO Agent MCP is a lightweight MCP server that connects to a single Watson Orchestrate agent. It exposes simple actions to invoke that agent and retrieve its details, making it ideal when you want a focused MCP integration without tool management or multiple agents.

How to use

You run the MCP server locally and connect to it from your MCP clients (Cursor, VS Code Copilot, Langflow, or similar). The server exposes two primary functions you will use through an MCP client: invoke_agent to send a message to the configured Watson Orchestrate agent, and get_agent to fetch details about that agent (name, description, tools, and instructions). You configure the agent once via environment variables or an MCP config, then your MCP clients can invoke the agent directly without managing tools or flows.

How to install

Prerequisites you need before installation: a Node.js runtime and npm. You will also require access credentials for Watson Orchestrate (WO_API_KEY, WO_INSTANCE_URL) and the target agent ID (WO_AGENT_ID). You will install the MCP package, build it, and run the server with your environment variables in place.

Step-by-step installation and startup:

Step-by-step commands

npm install
npm run build
cp .env.example .env
# Edit .env with your credentials and agent ID
WO_API_KEY=your-api-key WO_INSTANCE_URL=https://your-instance-id.orchestrate.ibm.com WO_AGENT_ID=your-agent-id node dist/index.js

Verify connectivity

Confirm that you can reach the configured agent by calling the two primary actions through your MCP client. You should be able to invoke the agent to respond to a message, and you should be able to retrieve the agent details with get_agent.

Test and troubleshooting

If you encounter issues, verify that your WO_API_KEY, WO_INSTANCE_URL, and WO_AGENT_ID are correctly set in your environment. Ensure the server started successfully and is listening on the expected interface. Use your MCP client to perform a basic invoke_agent call and then call get_agent to verify the agent’s metadata and available tools.

Configuration

Configure the MCP client or environment to point to the WxO Agent MCP server and the Watson Orchestrate agent. The server operates with a single agent and does not manage multiple agents or flows.

Security and data handling

Keep your WO_API_KEY and WO_AGENT_ID secure. Do not expose these credentials in public or shared environments. Use environment-level protections and rotate credentials as recommended by your security policy.

Notes and concepts

This MCP server is designed to be lightweight and purpose-built for a single Watson Orchestrate agent. It does not provide tool management, agent listing, or flow orchestration. It focuses on two capabilities: invoking the configured agent and retrieving its details.

VS Code and Langflow integration

Set up a Copilot Chat or Langflow integration by registering the WxO Agent MCP server as an MCP endpoint in your environment. Load your .env file in the integration workspace and use invoke_agent for direct agent queries. For quick prompts, you can ask the agent what it can help you with or request specific tasks, keeping interactions focused on the single configured agent.

Available tools

invoke_agent

Send a message to the configured Watson Orchestrate agent. The agent responds using its tools and LLM.

get_agent

Get details of the configured agent, including its name, description, tools, and instructions.