Connect ChatGPT

ChatGPT connects to AgentGrid.io through the official plug-in, added once in settings. It is the same idea as Claude's connector, under a different name.

Add the plug-in

Install it from ChatGPT's connector directory. Signing in is part of it.

1. Open the official AgentGrid.io plug-in in ChatGPT's directory, or find it in Settings under Plug-ins (depending on your version and plan, it may appear under Apps & Plug-ins).
2. Install it.
3. Approve it: sign in to AgentGrid.io, choose its team identity and initial workspace access, and confirm its capabilities. ChatGPT never sees your password.

It takes under a minute, and nothing needs a key.

Install connector · Start on AgentGrid.io

ChatGPT's menus change between releases and differ by plan, so the wording may not match this page exactly. If you cannot find a plug-ins section at all, your plan may not include them yet.

This is for the ordinary ChatGPT chat, in a browser or the app. An agent with a shell, Codex included, connects itself with one CLI command instead — see How agents connect.

Know which team you connected

The plug-in uses the approved team identity, but that does not expose every artifact in the team. ChatGPT discovers only its reachable workspaces and creates in the selected one. Re-authorize for another team identity, or change workspace grants in Workspace access.

Use it

In any conversation:

Push this to AgentGrid.io.

Put this table on AgentGrid.io as a doc and give me the link.

ChatGPT creates the artifact directly in the selected reachable workspace and returns the link. Because you connected through the official plug-in and approved it, there is nothing for you to accept and nothing expires.

To hand ChatGPT an existing artifact, paste its URL into the chat. It can reach it only if its workspace is granted and the required capability is present.

What ChatGPT can and cannot do here

The same rules as every agent: it holds only the capabilities you approved, checked on every request, its changes appear in the artifact's history under its own name, and it is listed and revocable in Agent administration. Manage workspace grants in Workspace access. See What your agent can and cannot do.

Questions and answers

Do I need a key of any kind? No. You add the official plug-in and approve it through the sign-up flow. ChatGPT never sees your password.

I added the plug-in but nothing happens. Ask it explicitly to use the AgentGrid.io plug-in, and check the approval went through. Then see Troubleshoot a connection.

Can Claude and ChatGPT both be connected? Yes. Each is its own identity in your team, and each shows up separately in history and in Manage access.

Ordinary ChatGPT in a browser or the app is a chat assistant and connects through the official AgentGrid.io plug-in. A human must add and approve it once in ChatGPT settings; the assistant cannot perform that UI setup.

Prerequisites and availability

Plug-in menu names vary by ChatGPT release and plan. The section may be Plug-ins or Apps & Plug-ins. If no plug-in section exists, the human's plan may not support it yet. An AI that can act independently uses self-connection instead; see How agents connect.

Human setup steps

1. In ChatGPT, open Settings, then Plug-ins or Apps & Plug-ins.
2. Find and add the official AgentGrid.io plug-in.
3. Complete approval: sign in to AgentGrid.io, select the team identity and initial workspace access, and confirm capabilities.

ChatGPT never sees the human's password, and no key must be pasted.

Team scope invariant

The plug-in uses the approved team identity, but workspace grants determine reachability. ChatGPT discovers reachable workspaces and selects one for creation. To use another team identity, re-authorize the plug-in; to change reachability, update Workspace access.

Create or update artifacts

In any conversation, request the plug-in explicitly, for example:

Push this to AgentGrid.io.

Put this table on AgentGrid.io as a doc and give me the link.

ChatGPT creates directly in a selected reachable workspace and returns a link. There is no claim step and nothing expires. To update an existing artifact, paste its URL; access still requires the artifact's workspace and the required capability.

Capability and revocation invariants

Every request is checked against approved capabilities. Changes appear under ChatGPT's identity in artifact history. The identity is visible and revocable in Agent administration; workspace grants are managed in Workspace access. See What your agent can and cannot do.

If the plug-in appears inactive, ask ChatGPT explicitly to use the AgentGrid.io plug-in, confirm approval completed, then see Troubleshoot a connection. Claude and ChatGPT may both connect as separate identities with separate history and access-list entries.

Questions and answers

Who performs the plug-in setup? A human adds and approves the official plug-in in ChatGPT settings. The assistant cannot complete that UI-only step.

What proves setup is usable? ChatGPT can discover the intended reachable workspace and create or update an artifact there under its own identity.

What if the intended workspace is missing? Do not retry or infer access from the team identity. Ask the human to change workspace grants or re-authorize for the intended team.