Connect Claude, ChatGPT or Cursor to Exoconnect

Connect the AI agent you already use to Exoconnect through MCP, on every plan. Sign in once, then choose what it may do on its own.

Updated · 4 min read

If you already work in Claude, ChatGPT, Cursor or another AI agent, you can connect it to Exoconnect. Then you can ask it, in the chat you already use, to answer comments, schedule posts, tag contacts or build an automation. It works in your workspace with your permissions, and you choose what it may do on its own.

This works on every plan, Free included.

How it works

Agents connect through MCP, the Model Context Protocol. It is an open standard that lets AI agents use other apps. Exoconnect runs an MCP server, and your agent talks to it.

  • You sign in, the agent doesn’t get a password. When you connect, the agent opens an Exoconnect sign-in page. You sign in and approve it. There’s no API key to copy or paste.
  • It can only do what you can. The agent acts as you. If your role in a workspace can’t publish, neither can your agent. If you lose access, so does the agent, on its very next step.
  • It sees your workspaces. If you belong to several, the agent can work in any of them you can open. Tell it which one you mean.

Add an agent

  1. In Exoconnect, go to Settings, then Agents.
  2. Choose Add agent.
  3. Pick your agent from the list, and follow its steps.

The list has Claude, ChatGPT, Claude Code, Codex, Cursor, VS Code (with GitHub Copilot) and Gemini CLI, plus Another agent for anything that supports MCP. Each guide shows the server address, with a button to copy it.

Here’s what each one takes, in short:

AgentWhat you do
Claude (claude.ai and the Claude apps)In Claude, open Customize, then Connectors. Choose + Add, then Add custom connector. Name it Exoconnect, paste the address, choose Continue, then sign in.
ChatGPTOpen chatgpt.com/plugins, choose +, then Add custom MCP server. Paste the address, choose OAuth, create it, then sign in.
Claude CodeRun the command shown in the guide in your terminal, then run /mcp in Claude Code and sign in.
CodexRun the two commands shown: one adds Exoconnect, the other signs you in. It then works in the Codex app, command line and editor extension.
CursorChoose Add to Cursor, confirm in Cursor, then sign in from Customize, MCPs.
VS CodeChoose Add to VS Code, confirm, and sign in when VS Code asks.
Gemini CLIRun the command shown, then /mcp auth exoconnect and sign in.
Another agentAdd a remote MCP server (Streamable HTTP) with the address, and sign in when it asks.

The address is https://api.exoconnect.in/mcp.

Two notes from the guides:

  • Claude on a Team or Enterprise plan: an owner first adds Exoconnect under Organization settings, Connectors. You then connect it yourself.
  • ChatGPT: custom servers depend on your ChatGPT plan, and in a ChatGPT workspace an admin may need to allow them.

Check it worked

Once connected, your agent shows up in Settings, then Agents, with the date you connected it. Try a simple request in your agent, such as:

  • “Which of my Exoconnect accounts have messages waiting for a reply?”
  • “Show me last month’s best Instagram posts.”

The agent asks Exoconnect what it can do, then reads what it needs. Reading never changes anything, so your agent can look freely.

Choose what it may do

Each agent has its own permission mode, shown on its row in Settings, then Agents. A new agent starts in Auto.

ModeWhat the agent may do
RestrictedCan look, can’t change anything
SupervisedAsks before every change
AutoMakes changes in Exoconnect, asks before anything reaches outside
Full accessActs without asking

“Asks” means the agent saves a draft instead of acting. You get a notification, and the draft waits for you under Drafts in the Assistant’s sidebar. Choose Approve to run it, or Reject. Drafts expire after 7 days.

Only you can change the mode, in Exoconnect. An agent can’t raise its own mode, and it can’t approve its own drafts.

One rule is stricter for outside agents: new posts always wait for your approval, in every mode. A few other actions always wait too, for every agent, such as publishing a form. What AI agents can do has the full picture.

One difference from Exi: when you decide a draft, your outside agent isn’t told. Ask it to check again, and it will see the result.

Disconnect an agent

In Settings, then Agents, choose Disconnect on the agent’s row, then confirm. Exoconnect cancels its access, and its next request is refused. To use it again, connect it again; it starts in Auto.

Connecting an agent, changing its mode and disconnecting it are kept in your own account history. Everything the agent does in a workspace is in that workspace’s history, under Settings, then Activity, with the agent’s name beside yours.

What it costs

Connecting outside agents is free on every plan. Your agent’s own AI is paid to its maker, as usual.

Some Exoconnect work costs money whoever asks for it, such as researching competitors’ accounts or analysing a video. That uses Exoconnect usage credits, which Pro and Team include. Automation runs count toward your plan’s monthly allowance. See the pricing page.

Exi or your own agent?

You can use both. Exi lives inside Exoconnect, needs no setup, and has its own computer for heavier work. Your own agent is handy when you already work there all day, or want Exoconnect next to your other tools in one chat. Both follow the same rules and the same permission modes.

Try it in your own account

Get started