ibm-baw

This MCP server exposes the REST services provided by your IBM Business Automation Workflow environment as tools that AI agents can invoke via MCP. Exposed REST services from workflows (such as claims handling or other business processes) are automatically surfaced as MCP tools, with the default version of each service used when multiple versions exist. The server communicates with MCP clients over stdio, making it suitable for local testing with IBM watsonx Orchestrate or other MCP clients. To get started, configure the ibm-baw MCP server with your workflow endpoint and credentials, then run the server. Clients can discover available tools and invoke them by name, passing the appropriate arguments defined by each REST service.

Using the server in practice typically involves: (1) setting up a connection in your MCP client to point at the local server, (2) authenticating using the configured credentials or Zen API key, and (3) invoking exposed tools that correspond to your workflow REST services. Since tools map directly to REST APIs exposed in your workflow environment, you can leverage natural language interfaces to trigger specific workflow actions (e.g., submitting a request, querying status, or performing a business operation) without implementing custom adapters.

Prerequisites:

• Python 3.12+ installed on your machine
• Network access from the MCP server to your IBM BAW environment
• Access credentials for a workflow user and, optionally, a Zen API key

Step 1: Clone or download the MCP server repository

• git clone https://github.com/your-org/ibmbpm-ibm-baw-mcp-server.git
• cd ibmbpm-ibm-baw-mcp-server

Step 2: Set up a Python virtual environment

• python -m venv venv
• source venv/bin/activate # On Windows: venv\Scripts\activate

Step 3: Install dependencies

• pip install -r requirements.txt

Step 4: Configure environment variables

• Create a .env file or export variables as shown in the mcp_config example. Key variables include ENDPOINT, VERIFY_SSL, USERID, PASSWORD, and optionally API_KEY and LOG_LEVEL.

Step 5: Run the MCP server

• python -m ibm_baw_mcp_server

Step 6: Verify operation

• Ensure the server starts without errors and check logs for successful tool registration. Use your MCP client to discover and invoke tools exposed by the server.

Tips and common considerations:

• Ensure your ENDPOINT is reachable from the machine running the MCP server; allowlist the server if your workflow environment requires it.
• If you are using Zen API keys, prefix the key as ZenApiKey<your-key> as described in the configuration guidance.
• Adjust VERIFY_SSL to false only in trusted environments with self-signed certificates to avoid SSL errors.
• The MCP server automatically exposes all REST services configured in your workflow environment; if you add new services, they become available as tools after restarting the server.
• When multiple versions exist for a REST service, only the default (currently active) version is exposed as a tool.
• For debugging, set LOG_LEVEL to DEBUG to get more granular logs.