Skip to main content
OneTest exposes MCP (Model Context Protocol) endpoints so AI coding assistants — such as Cursor, Windsurf, Claude Code, or any MCP-compatible tool — can interact with OneTest services directly from your editor.

​
Available Endpoints

Each OneTest service has its own MCP endpoint at /mcp/{service-name}:
ServiceEndpoint
Test Managementhttps://tms.onetest.ai/mcp/test-management
Artifactshttps://tms.onetest.ai/mcp/artifacts
Membershiphttps://tms.onetest.ai/mcp/membership
Meteringhttps://tms.onetest.ai/mcp/metering
Credential Poolhttps://tms.onetest.ai/mcp/credpool

​
Authentication

All MCP endpoints authenticate via API key using a Bearer token. Include your OneTest API key in the Authorization header: 
Authorization: Bearer ak_YOUR_API_KEY
To get an API key, go to Settings > API Keys.

​
Setup

Run the QA Agent init command and follow the browser-connect flow. It detects your editor, authenticates you, and writes the MCP configuration automatically: 
npx github:onetest-ai/qa-agent init

​
Option 2: Manual .mcp.json

Add OneTest endpoints directly to your .mcp.json configuration file. The example below connects to the Test Management service: 
{
  "mcpServers": {
    "onetest": {
      "url": "https://tms.onetest.ai/mcp/test-management",
      "headers": {
        "Authorization": "Bearer ak_YOUR_API_KEY"
      }
    }
  }
}
Add additional entries for other services (artifacts, credpool, etc.) as needed.
The npx github:onetest-ai/qa-agent init flow is the easiest way to get started — it handles authentication and writes the correct .mcp.json entries for all services.

​
What’s Next?

AI Assistant

Use the AI with your connected tools

LLM Configuration

Configure the AI provider powering the assistant