Use ProjectHQ from ChatGPT, Claude or Cursor

Connect ChatGPT, Claude, Claude Code or Cursor to your ProjectHQ project so you can ask about your rankings, traffic and content, and let it write drafts.

Connect ProjectHQ to the AI assistant you already use — ChatGPT, Claude, Claude Code, Cursor or any other app that supports MCP (Model Context Protocol). Then ask it about your project in plain words: "Which pages lost traffic this week?", "What did visitors ask in the chat today?", "Draft an article on the keyword we're closest to page one for." If you allow it, the assistant can also write drafts for you.

What your assistant can do

Every connection can read your project:

  • Analytics — visitors, traffic over time, top pages and sources, live visitors, notable spikes and drops, events and goals.
  • Search — your keyword rankings, Google Search Console clicks and queries, indexing status, and Bing data.
  • Content — your articles, publications, content plan and documentation.
  • SEO — site audit issues, insights and your webpages.
  • Leads — contacts, chat conversations and your Knowledge Base.
  • Notifications — what happened in your project lately.

With Allow editing content ticked, it can also create and edit drafts of articles and documentation pages, arrange documentation pages, and upload images into them.

It can't publish, delete anything or change your settings. Drafts it writes wait for you in Publications until you review and publish them; editing an already published page doesn't change the live version until you publish the edit.

Note: A connected assistant can read your contacts and conversations. Connect only assistants and accounts you trust with that data.

Connect your assistant

  1. In ProjectHQ, click Settings at the bottom of the menu.
  2. Under Account, open AI assistants.
  3. Under Connect an AI assistant, copy the server URL with the copy button. The address is the same for everyone: 
    https://api.projecthq.pro/v1/mcp
  4. Pick your assistant's tab and follow the steps (also below).
  5. When your assistant opens the ProjectHQ sign-in, sign in and approve the connection (next section).
The AI assistants settings: Connect an AI assistant with the server URL and tabs for ChatGPT, Claude, Claude Code and Cursor, Connected assistants and API key

ChatGPT

  1. Open Settings → Apps & Connectors → Advanced settings and turn on Developer mode.
  2. Go back to Connectors and click Create (or Add custom connector).
  3. Paste the server URL and choose OAuth authentication.
  4. Click Connect, sign in to ProjectHQ and pick the project.

Claude (web and desktop)

  1. Open Settings → Connectors → Add custom connector.
  2. Paste the server URL and confirm.
  3. Click Connect — a ProjectHQ window opens where you sign in and pick a project.

Claude Code

Run this in your terminal:

claude mcp add --transport http projecthq https://api.projecthq.pro/v1/mcp

Then run /mcp inside Claude Code, choose projecthq and Authenticate.

Cursor

Add this to .cursor/mcp.json in your project (or the global one in ~/.cursor):

{
  "mcpServers": {
    "projecthq": {
      "url": "https://api.projecthq.pro/v1/mcp"
    }
  }
}

Then click Connect next to the server in Cursor Settings → MCP.

The menus in ChatGPT, Claude and Cursor change from time to time. If a step doesn't match, look for "connectors", "custom connector" or "MCP servers" in your assistant's settings.

Approve the connection

When your assistant sends you to ProjectHQ, you see Connect … to ProjectHQ:

The approval screen: Connect Claude to ProjectHQ, what it can read, Allow editing content, Choose a project, and the Cancel and Authorize buttons
  1. Check what it will be able to read.
  2. Tick Allow editing content if you want it to write drafts and upload images. Leave it unticked for read-only access.
  3. Under Choose a project, click the project to connect. One connection covers one project; connect again to add another.
  4. Click Authorize.

You're sent back to your assistant, and it's ready. The approval link works once and expires after 15 minutes — if you see This request is no longer valid, start the connection again from your assistant.

To switch a read-only connection to editing (or back), disconnect it and connect again with the box ticked or unticked.

See and disconnect assistants

Settings → AI assistants → Connected assistants lists every assistant connected to the project, who connected it, when, and when it was last used. Each shows Read only or Can edit drafts. Click Disconnect to cut it off — it takes effect immediately.

Use an API key instead (advanced)

For scripts and tools that can't sign in, use an API key. Under API key, click Generate API key and copy it straight away: it's shown only once. Send it as a bearer token in the Authorization header to the same server URL.

An API key gives read access to this project only — editing always needs a signed-in connection. Treat the key like a password. Regenerate replaces it with a new one; Revoke turns it off. Assistants connected by signing in aren't affected by either.

Use ProjectHQ from ChatGPT, Claude or Cursor — ProjectHQ Docs