Summary
One tool, and its parameters are the whole interface.
You still write the Locust file; the server runs it with the user count, spawn rate, runtime and target host you give, headless by default or with the UI when you want to watch. The value lands on the way back — the output arrives in the conversation, so the analysis happens where the numbers are.
What it is
A Python server that runs your Locust test script with the parameters you name and streams the output back — headless or with the UI, against the host, runtime, user count and spawn rate you choose.
What you get
- `run_locust` — takes `test_file`, plus `headless`, `host`, `runtime`, `users` and `spawn_rate`
- Defaults set once in a `.env`: `LOCUST_HOST`, `LOCUST_USERS`, `LOCUST_SPAWN_RATE`, `LOCUST_RUN_TIME`
- Real-time output from the run rather than a summary after the fact
- HTTP and HTTPS support out of the box, and your own custom task scenarios
Requirements
Python 3.13 or higher and the uv package manager. Clone the repo, `uv pip install -r requirements.txt`, then point your client at `uv --directory /path/to/locust-mcp-server run locust_server.py`. You supply the Locust script — the tool takes its path.