AI connectors

Use Axonplane from the AI apps you already use

Connect Claude, ChatGPT, Claude Code, Cursor or VS Code, and ask about your projects, pipeline, support queue and team in plain language. It acts as you, and only where you allow.

The address to give your app

https://api.axonplane.com/api/v1/mcp

Part of the Pro and Enterprise plans · Uses the open Model Context Protocol

Connect in a minute

Give the app the address above. It sends you to Axonplane, where you pick a workspace and what the app may reach — then you are back in the app, connected.

  1. 1

    Claude — web, desktop and phone

    Open Settings → Connectors, choose Add custom connector, paste the address and connect. Claude sends you here to choose what to allow.

  2. 2

    ChatGPT

    Turn on developer mode in ChatGPT’s connector settings, then create a connector with the address, signing in with OAuth.

  3. 3

    Claude Code

    Run claude mcp add --transport http axonplane https://api.axonplane.com/api/v1/mcp, then /mcp inside Claude Code and choose Authenticate.

  4. 4

    Cursor and VS Code

    Add the address as an HTTP MCP server in mcp.json. Each asks you to sign in the first time it connects.

  5. 5

    Anything else

    Any client that speaks MCP over Streamable HTTP can sign in the same way, or use an API key from Settings → Connect AI as a bearer token.

Ask it about your whole company

38 tools across work, customers, support, chat and people — reading, and where you allow it, doing.

Projects and tasks

  • What’s overdue in the launch project?
  • Create a task for Priya to review the copy, due Friday
  • Move the onboarding tasks to Done

CRM

  • Add Acme and its CTO, and open a $40k deal for them
  • Which deals are expected to close this month?
  • Log the call I just had with Jordan

Support desk

  • Summarise the urgent tickets from this week
  • Draft an internal note on ticket 482
  • Open a ticket for the invoice problem Sam phoned about

Team chat

  • What did #engineering decide about the release?
  • Post a summary of today’s progress in #launch

Notifications

  • What needs my attention today?

HR — Enterprise

  • Request next Friday off
  • Who is in the design team, and who manages it?
  • Which roles are we hiring for?

You stay in control

An AI app gets exactly what you and your admins allow, and you can see everything it did.

You choose what it can reach

When you connect an app you pick one workspace and, area by area, whether it may read or also make changes. It never gets more than you have.

Admins set the ceiling

Owners and admins can switch AI apps off for the whole workspace, keep them to admins, or limit any area — applied on the next call, to every app already connected.

Nothing destructive

No tool deletes anything, exports data, emails a customer or touches payroll. Replies to support tickets are internal notes unless you ask for one to reach the customer.

Disconnect in one click

Settings → Connect AI lists every connected app, and disconnecting one stops it at once. Every call it made is in the audit log.

Included in Pro and Enterprise

Pro

  • 5,000 tool calls a month
  • Projects, tasks, CRM, support, chat and notifications
  • Sign-in for Claude, ChatGPT, Cursor and more
  • Admin rules and one-click disconnect

Enterprise

  • 50,000 tool calls a month
  • Everything in Pro
  • HR tools: directory, time off, open roles
  • Export the AI audit trail as CSV

Questions

Do I need an API key?
No. Claude, ChatGPT, Claude Code, Cursor and VS Code sign in: you paste the address, the app sends you to Axonplane, and you choose what to allow. API keys remain for scripts and for apps that cannot sign in.
What does the app see?
Only what it reads through the tools you allowed, as you — so never more than you could see yourself. How the app itself treats that data is governed by that app’s own terms.
What happens at the monthly limit?
Owners and admins are emailed at 80% and again at the limit. After that, calls are refused until the allowance resets on the first of the month; nothing else in Axonplane is affected.
Can someone trick the assistant through a support ticket?
Tickets arrive from your public portal, so their text is written by strangers. Axonplane tells the assistant to treat everything it reads as data, not instructions, and offers no tool that sends data out of the workspace — but review what an assistant proposes after it has read one.

Already have a workspace? Everything is at Settings → Connect AI.