Fastweigh MCP

This is an MCP server that exposes the Fastweigh GraphQL API to AI clients (Claude, ChatGPT, Microsoft Copilot, Gemini, Cursor, VS Code, and anything else that speaks the Model Context Protocol).

Every client follows the same pattern: point it at https://mcp.fastweigh.com/mcp, and on first use you land on a consent screen where you paste your Fastweigh API key and choose read-only or read/write access.

Add to Claude

Add to Claude (web, desktop, mobile & Claude Code)

The button above opens claude.ai with the connector pre-filled. To add it by hand:

  1. Open Settings → Connectors (in the "Customize Claude" section) and click Add custom connector.
  2. Enter https://mcp.fastweigh.com/mcp as the URL and click Add. Leave the advanced OAuth fields blank — the server registers itself.
  3. Click Connect, then sign in with your Fastweigh API key and pick scopes.
  4. In a chat, open the tools menu and enable Fastweigh.

Connectors run from Anthropic's cloud, so one setup covers claude.ai, Claude Desktop, and the mobile apps. On Team / Enterprise plans, an owner can add Fastweigh once under Organization settings → Connectors; each member then connects with their own key.

Claude Code

claude mcp add --transport http fastweigh https://mcp.fastweigh.com/mcp

Add --scope user to make it available in every project. Then run /mcp inside Claude Code to complete the sign-in.

Add to ChatGPT

ChatGPT doesn't have a one-click connector link. How you add Fastweigh depends on your plan.

Business / Enterprise / Edu (recommended)

A workspace admin or owner adds Fastweigh once and publishes it to the workspace. Members then use it without enabling developer mode themselves.

  1. As an admin, open Workspace settings → Apps → + Create.
  2. Fill in the form:
    • Icon (optional): download fwicon-128.png (128×128, transparent).
    • Name: Fastweigh
    • Description: Query and update your Fastweigh tenant via GraphQL.
    • MCP Server URL: https://mcp.fastweigh.com/mcp
    • Authentication: OAuth (settings auto-discover from the URL)
  3. Confirm the risk acknowledgement, create, then publish to the workspace.
  4. Each member signs in once by pasting their own Fastweigh API key and granting scopes.

Plus / Pro (individual)

Individual accounts can only add custom MCP servers with developer mode enabled, and ChatGPT shows a prominent "unreviewed" warning for servers added this way.

  1. Enable developer mode: Settings → Apps → Advanced settings → Developer mode.
  2. Back in Settings → Apps, click Create and fill in the same fields as above.
  3. Acknowledge the warning and create.
  4. Sign in with your Fastweigh API key when prompted.

On older builds the same screens live under Settings → Connectors.

Add to Microsoft Copilot (Microsoft 365)

Microsoft 365 Copilot itself doesn't have a UI for adding MCP servers. Fastweigh is added through Copilot Studio, then published into Copilot Chat for your tenant. A maker (admin or builder) does this once; everyone else just talks to the agent.

Prerequisites

Add Fastweigh as a tool in Copilot Studio

  1. Open copilotstudio.microsoft.com and sign in with your work account.
  2. Create a new agent (e.g. Fastweigh) or open the one you want to extend.
  3. Go to Tools → + Add a tool → New tool → Model Context Protocol.
  4. Fill in the form:
    • Server name: Fastweigh
    • Description: Query and update your Fastweigh tenant via GraphQL.
    • Server URL: https://mcp.fastweigh.com/mcp
    • Authentication: OAuth 2.0 — Copilot Studio auto-discovers the endpoints. Leave client ID / secret blank; the server uses Dynamic Client Registration.
  5. Save. Copilot Studio creates a custom connector behind the scenes wrapping the MCP server, then registers the three Fastweigh tools (search_schema, execute, execute_mutation) on your agent.
  6. The first time you (or any user) invoke the agent, you'll be redirected to consent: paste your own Fastweigh API key and pick scopes (read-only, or read/write).

Publish to Microsoft 365 Copilot Chat

  1. In your agent, open Channels → Microsoft 365 Copilot Chat → Turn on.
  2. Hit Publish, then Submit for admin approval if your tenant requires it.
  3. Once approved, users can mention the agent in M365 Copilot Chat (e.g. @Fastweigh what's our inbound tonnage this week?) and Copilot will call Fastweigh on their behalf.

Don't want to deal with OAuth?

You can register the tool with API key auth instead of OAuth and paste a Fastweigh tenant key directly in the connector settings. Every user of the agent will share that key, so use a scoped read-only key unless you've thought through the blast radius.

Add to Gemini

Gemini app (gemini.google.com)

Custom MCP servers connect through Connected apps for Gemini Spark.

  1. On a computer, open gemini.google.com, then Settings & help → Connected apps.
  2. Under Custom apps for Spark, click Add a custom app.
  3. Enter https://mcp.fastweigh.com/mcp as the MCP server URL and click Next. No client credentials are needed — the server registers itself.
  4. Sign in with your Fastweigh API key on the consent screen.

Requires Gemini Spark on a personal Google account (US-only at the time of writing). Connect on the web; the app then works in the Gemini mobile app too.

Gemini CLI

gemini mcp add --transport http fastweigh https://mcp.fastweigh.com/mcp

This adds it to the current project; use -s user for all projects. Run /mcp auth fastweigh in the CLI to sign in.

Add to Cursor

One-click install (opens Cursor), or add it by hand under Cursor Settings → MCP → New MCP server, which edits ~/.cursor/mcp.json:

{
  "mcpServers": {
    "fastweigh": { "url": "https://mcp.fastweigh.com/mcp" }
  }
}

Cursor prompts you to authenticate the server the first time it connects — sign in with your Fastweigh API key.

Add to VS Code (GitHub Copilot)

One-click install (opens VS Code), or from a terminal:

code --add-mcp '{"name":"fastweigh","type":"http","url":"https://mcp.fastweigh.com/mcp"}'

Or run MCP: Add Server from the Command Palette and choose HTTP. VS Code opens a browser window to sign in the first time the server is used.

Other clients, scripts & automation

The server speaks standard MCP (Streamable HTTP) at https://mcp.fastweigh.com/mcp, with OAuth 2.1, PKCE, and Dynamic Client Registration. Any spec-compliant client connects with just the URL:

{
  "mcpServers": {
    "fastweigh": { "url": "https://mcp.fastweigh.com/mcp" }
  }
}

Scripts and CI (no OAuth)

For unattended use, skip the OAuth flow by sending your Fastweigh API key directly as a bearer token:

{
  "mcpServers": {
    "fastweigh": {
      "url": "https://mcp.fastweigh.com/mcp",
      "headers": { "Authorization": "Bearer <your-fastweigh-api-key>" }
    }
  }
}

Scopes follow the key: a read-only key gets queries, a read/write key also gets mutations. Prefer a read-only key for automation.

Clients without remote-server support

Bridge through mcp-remote: configure a stdio server with command npx and args ["-y", "mcp-remote", "https://mcp.fastweigh.com/mcp"].

Testing

Point MCP Inspector (npx @modelcontextprotocol/inspector) at the URL to browse the tools and debug.

How it works

  1. Your AI client opens a connection to this server and authorizes via OAuth.
  2. On the consent screen you paste your Fastweigh API key and pick the scopes to grant (fastweigh:read, optionally fastweigh:readwrite).
  3. The client gets a short-lived access token; your Fastweigh key stays on our servers, encrypted at rest.
  4. When the AI wants to query Fastweigh, it writes small snippets of JavaScript. Those run in an isolated sandbox with no outbound network access except the Fastweigh GraphQL endpoint, authenticated with your key.
  5. Every mutation is approved by you in chat before it runs.

Tools the AI sees

Getting an API key

Contact your Fastweigh administrator to issue a per-tenant API key. Pick read-only if you only need the AI to read.