TestRail MCP Server

This Model Context Protocol (MCP) server provides tools for interacting with TestRail directly from Claude AI and other MCP-supported clients like Cursor. It allows you to manage test cases, projects, suites, runs, and more without leaving your conversation with the AI.

Available Tools

The TestRail MCP server provides the following tools:

Usage

You can connect this MCP server by setting like the below. This method uses npx to automatically download and run the latest version of the package, eliminating the need for local installation.

// Example configuration using npx
{
  "mcpServers": {
    "testrail": {
      "command": "npx",
      "args": ["@bun913/mcp-testrail@latest"],
      "env": {
        "TESTRAIL_URL": "https://your-instance.testrail.io", // Replace with your TestRail URL
        "TESTRAIL_USERNAME": "your-email@example.com", // Replace with your TestRail username
        "TESTRAIL_API_KEY": "YOUR_API_KEY" // Replace with your TestRail API key
      }
    }
  }
}

Troubleshooting

• spawn npx ENOENT / spawn node ENOENT (commonly on macOS): your MCP host (Cursor, Claude Code, Claude Desktop, …) cannot find npx or node at process-spawn time. The chat UI usually surfaces this as a generic "MCP server doesn't work" with no useful detail; the per-server log is the diagnostic source of truth.

  Why it happens on macOS: GUI apps launched from the Dock, Spotlight, or Finder inherit launchd's minimal PATH (/usr/bin:/bin:/usr/sbin:/sbin). If you installed Node via a version manager (nvm, asdf, mise, fnm, Volta) or Apple Silicon Homebrew (/opt/homebrew/bin/), npx lives outside that PATH — only your shell startup file (~/.zshrc / ~/.bashrc) adds it. Your terminal works because the shell ran the startup file; the GUI-app process never did.

  Diagnose by checking the per-server log for spawn npx ENOENT:

  • Cursor: ~/Library/Application Support/Cursor/logs/<session>/window<N>/exthost/anysphere.cursor-mcp/MCP <server>.log
  • Claude Code / Claude Desktop: ~/Library/Logs/Claude/

  Fix by replacing "npx" in your MCP config with its absolute path. Run which npx in your normal terminal:

  /Users/you/.nvm/versions/node/v24.15.0/bin/npx   # nvm
  /opt/homebrew/bin/npx                            # Apple Silicon Homebrew
  /usr/local/bin/npx                               # Intel Homebrew / system Node
  

  Then update your MCP config:

  {
    "mcpServers": {
      "testrail": {
        "command": "/Users/you/.nvm/versions/node/v24.15.0/bin/npx",
        "args": ["@bun913/mcp-testrail@latest"],
        "env": {
          "TESTRAIL_URL": "https://your-instance.testrail.io",
          "TESTRAIL_USERNAME": "your-email@example.com",
          "TESTRAIL_API_KEY": "YOUR_API_KEY"
        }
      }
    }
  }
  

  Restart your MCP client after the change. The same fix applies to every npx-launched MCP server — if mcp-testrail is failing for this reason, your other npx-launched servers are likely failing too.

• Authentication issues: Check your TestRail API credentials.

• Your conversation is too long: Use limit and offset parameters for test cases and sections to paginate results.

• HTTP 400 errors when creating/updating test cases: TestRail projects have different templates, custom fields, and required fields. This MCP server passes your parameters directly to the TestRail API — it does not validate or transform them. If you encounter 400 errors, define your project's rules in CLAUDE.md or AGENTS.md so the LLM sends the correct parameters. For example:

  # TestRail Rules for This Project
  - Project ID: 1
  - Always use template 2 (Separated Steps) when creating test cases
    - Use `customStepsSeparated` (array of step objects)
    - Do NOT send `customSteps` or `customExpected` with template 2
  - Required custom fields: custom_automation_type (default: 0)
  - Call `getCaseFields` at the start of a session to check available fields
  

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Acknowledgements

• TestRail API
• Model Context Protocol SDK