api-test-mcp

api-test-mcp is an interface testing MCP server that lets you create test plans, run many interface tests in batches, automatically evaluate results, and export test data for analysis. It helps you validate API functionality, robustness to bad input, and track testing outcomes over time.

How to use

You connect an MCP client to the server to manage test plans and run tests. Use an HTTP MCP endpoint to access the server remotely, or run the server locally and connect via its standard local API path. You create a test plan, execute batch test tasks, review automated evaluations, and export results for reporting.

Key usage patterns:

• Add a new test plan that describes each API endpoint, its expected inputs, and success criteria.
• Run a batch of test tasks to exercise multiple endpoints or scenarios in one go.
• Let the server automatically assess results to surface failures, performance issues, or invalid responses.
• Export results to share with teammates or plug into dashboards.

How to install

Prerequisites:

• Node.js (recommended) or a compatible runtime as described by the server build.
• Docker (optional for containerized deployment).
• Access to a web browser or MCP client to connect to the server endpoint.

# Local deployment via Docker
git clone https://github.com/Actrue/api-test-mcp.git
cd api-test-mcp

docker build -t api-test-mcp .

docker run -p 3000:3000 -v $(pwd)/data:/usr/src/app/data api-test-mcp

# Then connect your MCP client to http://localhost:3000/mcp

Additional notes

To start a local server using the standard command flow, the project uses a script that initiates the server on port 3000. The final runtime command to start the server, as shown in the local setup, is pnpm run go. Use this command in the local environment you prepared to start the MCP server, then point your MCP client to the URL http://localhost:3000/mcp.