Locust MCP Server

1# 🚀 ⚡️ locust-mcp-server
2 
3A Model Context Protocol (MCP) server implementation for running Locust load tests. This server enables seamless integration of Locust load testing capabilities with AI-powered development environments.
4 
5## ✨ Features
6 
7- Simple integration with Model Context Protocol framework
8- Support for headless and UI modes
9- Configurable test parameters (users, spawn rate, runtime)
10- Easy-to-use API for running Locust load tests
11- Real-time test execution output
12- HTTP/HTTPS protocol support out of the box
13- Custom task scenarios support
14 
15![Locust-MCP-Server](./images/locust-mcp.png)
16 
17## 🔧 Prerequisites
18 
19Before you begin, ensure you have the following installed:
20 
21- Python 3.13 or higher
22- uv package manager ([Installation guide](https://github.com/astral-sh/uv))
23 
24## 📦 Installation
25 
261. Clone the repository:
27 
28```bash
29git clone https://github.com/qainsights/locust-mcp-server.git
30```
31 
322. Install the required dependencies:
33 
34```bash
35uv pip install -r requirements.txt
36```
37 
383. Set up environment variables (optional):
39   Create a `.env` file in the project root:
40 
41```bash
42LOCUST_HOST=http://localhost:8089  # Default host for your tests
43LOCUST_USERS=3                     # Default number of users
44LOCUST_SPAWN_RATE=1               # Default user spawn rate
45LOCUST_RUN_TIME=10s               # Default test duration
46```
47 
48## 🚀 Getting Started
49 
501. Create a Locust test script (e.g., `hello.py`):
51 
52```python
53from locust import HttpUser, task, between
54 
55class QuickstartUser(HttpUser):
56    wait_time = between(1, 5)
57 
58    @task
59    def hello_world(self):
60        self.client.get("/hello")
61        self.client.get("/world")
62 
63    @task(3)
64    def view_items(self):
65        for item_id in range(10):
66            self.client.get(f"/item?id={item_id}", name="/item")
67            time.sleep(1)
68 
69    def on_start(self):
70        self.client.post("/login", json={"username":"foo", "password":"bar"})
71```
72 
732. Configure the MCP server using the below specs in your favorite MCP client (Claude Desktop, Cursor, Windsurf and more):
74 
75```json
76{
77  "mcpServers": {
78    "locust": {
79      "command": "/Users/naveenkumar/.local/bin/uv",
80      "args": [
81        "--directory",
82        "/Users/naveenkumar/Gits/locust-mcp-server",
83        "run",
84        "locust_server.py"
85      ]
86    }
87  }
88}
89```
90 
913. Now ask the LLM to run the test e.g. `run locust test for hello.py`. The Locust MCP server will use the following tool to start the test:
92 
93- `run_locust`: Run a test with configurable options for headless mode, host, runtime, users, and spawn rate
94 
95## 📝 API Reference
96 
97### Run Locust Test
98 
99```python
100run_locust(
101    test_file: str,
102    headless: bool = True,
103    host: str = "http://localhost:8089",
104    runtime: str = "10s",
105    users: int = 3,
106    spawn_rate: int = 1
107)
108```
109 
110Parameters:
111 
112- `test_file`: Path to your Locust test script
113- `headless`: Run in headless mode (True) or with UI (False)
114- `host`: Target host to load test
115- `runtime`: Test duration (e.g., "30s", "1m", "5m")
116- `users`: Number of concurrent users to simulate
117- `spawn_rate`: Rate at which users are spawned
118 
119## ✨ Use Cases
120 
121- LLM powered results analysis
122- Effective debugging with the help of LLM
123 
124## 🤝 Contributing
125 
126Contributions are welcome! Please feel free to submit a Pull Request.
127 
128## 📄 License
129 
130This project is licensed under the MIT License - see the LICENSE file for details.
131