IBM’s wxflows MCP example exposes the tools deployed in a watsonx.ai Flows Engine project through a local TypeScript server.
IBM’s wxflows MCP example exposes the tools deployed in a watsonx.ai Flows Engine project through a local TypeScript server.
The IBM repository includes a TypeScript MCP example that loads tools from a watsonx.ai Flows Engine project. The server obtains tool definitions from the wxflows SDK and delegates calls to the deployed backend. The sample project supplies Google Books and Wikipedia tools; another project's tool list follows its own deployment.
Follow the example to configure and deploy the Flows Engine project before starting the MCP server. Install the JavaScript dependencies and build the local server. Its stdio launch uses Node.js with the compiled build/index.js file. The example takes a project endpoint and API key through WXFLOWS_ENDPOINT and WXFLOWS_APIKEY.
The Flows Engine endpoint is the backend used by the SDK. It is not a universal hosted MCP URL that can replace the local bridge. Likewise, the repository's general SDKs and tools are broader than the runnable MCP example. The JavaScript directory contains the implemented server; the example documentation does not supply an equivalent Python server.
Confirm that tool discovery matches the project you deployed. Search for a known book or Wikipedia page and compare the requested fields with the returned data. Check how missing results and backend errors appear in the client. Add a custom tool only after defining its input schema, data access and output format in the Flows Engine project.
Check the linked publisher documentation and your connected client's tool list. Names and availability can vary with the server version and account permissions.
| Tool | What it does |
|---|---|
| google_books | Google Books search tool supplied by the sample Flows Engine deployment. |
| wikipedia | Wikipedia search and page tool supplied by the sample deployment. |
{
"mcpServers": {
"wxflows-server": {
"command": "node",
"args": [
"/ABSOLUTE_PATH/wxflows/examples/mcp/javascript/build/index.js"
],
"env": {
"WXFLOWS_APIKEY": "YOUR_WXFLOWS_APIKEY",
"WXFLOWS_ENDPOINT": "YOUR_WXFLOWS_ENDPOINT"
}
}
}
}Deploy the sample or your own project first, then install dependencies and build the JavaScript MCP example. Replace the absolute path and project credentials.
Node.js, a configured Flows Engine project, its endpoint and API key, and the built TypeScript MCP example.
| Variable | Description | Required |
|---|---|---|
| WXFLOWS_APIKEY | API key for the deployed Flows Engine project. | Yes |
| WXFLOWS_ENDPOINT | Endpoint for the deployed Flows Engine project. | Yes |
Stop letting your assistant hallucinate n8n node parameters — this server hands it the real schemas, templates and validation.
Enables AI assistants to interact with and manage n8n workflows through natural language.
Dify Workflows MCP exposes configured Dify applications as assistant tools through a local TypeScript server.
Integrate Linear project management with Claude for enhanced workflow automation.
AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs
Build and run agents and automations across hundreds of apps using natural language.