Trax

Model Context Protocol

Connect your AI assistant to Trax

Trax speaks MCP, the open standard that lets AI assistants talk to real tools. So Claude, Codex, Cursor, or whatever you use can read and update your rail project data right in a conversation. Point it at one URL and sign in. From there it works inside your own account, and it can't do anything you couldn't do yourself.

Trax MCP serverhttps://trax-mcp.res.app/mcp

It only sees what you can

Every request signs in as you and runs with your permissions, so the assistant sees and changes exactly what you could in the app, and nothing else. When you connect, you pick which projects the session can touch, and that choice is locked into the token it's given. Writes are limited to project-setup fields, and the assistant can't widen its own access without you approving it.

Connect your assistant

Pick your tool below. Claude connects on its own; everything else reaches the same endpoint throughmcp-remote, a small helper that opens a browser for sign-in the first time.

Claude (web & desktop)

Settings → Connectors → Add custom connector. Paste the URL and leave the client id/secret blank; Claude registers itself. Turn it on per chat from the + menu.

Remote MCP server URL:
https://trax-mcp.res.app/mcp

Claude Code (CLI)

Add the server once, then run /mcp inside Claude Code to sign in through your browser.

claude mcp add --transport http trax https://trax-mcp.res.app/mcp

Codex CLI

Codex talks to local (stdio) servers, so bridge to the remote endpoint with mcp-remote. Drop this in ~/.codex/config.toml. A browser handles sign-in the first time.

[mcp_servers.trax]
command = "npx"
args = ["-y", "mcp-remote", "https://trax-mcp.res.app/mcp"]

Cursor, Windsurf & other MCP clients

Any client that reads an mcp.json (or equivalent) connects through the same bridge.

{
  "mcpServers": {
    "trax": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://trax-mcp.res.app/mcp"]
    }
  }
}

What happens when you sign in

  1. 1

    Sign in with Trax

    A Trax sign-in page opens. Same email and password you use in the app.

  2. 2

    Choose your projects

    Then a project picker lets you grant the session access to the projects you want it to work in. You can re-authorise later to widen that, or let the assistant ask for more (you approve each time).

  3. 3

    Ask away

    A good first message is simply "call whoami" to confirm the connection, then ask it about your assets, documents or work logs.

What your assistant can do

Read your project data

Assets, cables, documents, media, work logs, issues, map layers and features, work instructions and templates. It can also read your schema: custom fields, asset types and value lists.

Set up and update projects

Create and update assets and cables, update map layers and features, attach media to assets, and raise work instructions from a template. Everything runs in bulk, one result per item, so a partial failure never loses the rest.

Answer questions from the register

Narrow thousands of assets down to the ones you mean, resolve map coordinates, measure route distances, or pull a photo inline. All of it without leaving the chat.

Don’t have Trax yet?

Talk to us about running your rail project’s QA, compliance and reporting in one live platform.

Get in touchExplore the product