How to drive T3 Code threads from an outside agent

T3 Code lets an agent it did not start work with an environment through that environment's MCP server. Those agents can read projects and threads, start and message threads, and check which providers and models are available. The outside-agents guide covers Claude Code or Codex in your own terminal, ChatGPT, and bots that support MCP. Each agent signs in once, and you choose what it may do.

Every client in the guide starts the same way: copy the environment MCP URL, sign in once, and choose what the agent may do. A hosted agent needs an address it can reach from the internet. An agent on your own computer can use a private address that machine already uses.

On October 7, 2026, Julius (@jullerino) wrote that the T3 Code MCP now works with any connector, and that you can add it to ChatGPT, Claude, Grok, and other connectors and drive your threads from there. That post confirms the client range. It is not a setup guide. It does not include a Grok command, and the docs do not either.

Copy the MCP URL

In Settings → Connections, open a saved environment's menu and choose Copy MCP URL. The URL is the environment's address followed by /mcp. The docs show this form:

https://<environment-address>/mcp

The copied URL uses the route this device is connected over, so an agent on the same device can reach it.

An agent on your own computers can use any address that computer reaches the environment at: a LAN or Tailscale address, T3 Connect, or localhost on the host. The outside-agents page points readers to a remote-access page for how those addresses are set up. Those steps are not in the pages used for this article, so they are not repeated here.

A hosted agent, such as ChatGPT or a bot running in the cloud, must reach the environment from the internet. Use the T3 Connect address. A Tailscale address only works from your own tailnet.

The first time an agent connects, it opens a sign-in page on the environment. The page names the agent and where its access goes.

For an agent on your computer, approval returns to a localhost address on the computer that opened the page.
For a hosted agent, approval goes to that service, such as chatgpt.com. Anyone who runs that service gets the access.

Only approve a sign-in you just started. To approve, enter a pairing code from Settings → Connections or from t3 auth pairing create on the host. A browser already signed in to the environment as an administrator can approve without a code.

Then choose what the agent may do:

Read only is the default. The agent can read projects and threads and see the available providers and models. It cannot change anything.
Supervised through Full access also let it start, message and stop threads in every project. A thread it starts or steers cannot run with more permissions than the mode you chose.

Leave the default if the outside agent should only inspect projects, threads, providers, and models. Pick a mode from Supervised through Full access when you want it to start, message and stop threads. The thread cannot run with more permissions than that mode.

Claude Code

The docs give these two commands. Use the MCP URL you copied in place of the placeholder address:

claude mcp add --transport http t3 https://<environment-address>/mcp
claude mcp login t3

claude mcp login opens the sign-in page in your browser. On a machine without a browser, add --no-browser, open the printed URL elsewhere, and paste the final URL back when asked. claude mcp list shows t3 as connected once you approve.

Codex

Codex uses a URL flag:

codex mcp add t3 --url https://<environment-address>/mcp
codex mcp login t3

codex mcp login --no-browser prints the sign-in URL instead of opening it. codex mcp list shows t3 with OAuth once you approve.

ChatGPT

Add the environment as a custom MCP app in ChatGPT's apps settings:

Create a new app or plugin with a custom MCP server.
Set the server URL to https://<environment-address>/mcp, using the T3 Connect address.
Choose OAuth for authentication. ChatGPT discovers the rest; you do not need a client ID or secret.
Create the app, then approve the sign-in page that opens.

ChatGPT is a hosted agent, so the address has to be reachable from the internet. Use the T3 Connect address. Do not give ChatGPT a Tailscale address or localhost; the docs say a hosted agent must use the T3 Connect address, and a Tailscale address only works from your own tailnet.

Other agents and bots

Any agent that supports remote MCP servers over HTTP with OAuth can connect. Give it the MCP URL and choose OAuth, and the agent finds the sign-in settings itself. Agents without OAuth support cannot connect.

That is also the path for Grok. Julius's post on October 7, 2026 names Grok alongside ChatGPT, Claude, and other connectors. The docs do not give a Grok-specific command. If the client can speak remote MCP over HTTP with OAuth, give it the same MCP URL and approve the sign-in. If it cannot, the docs say it cannot connect.

Remove an agent

Approved agents appear under Settings → Connections like other clients. Revoke one there to cut off its access immediately. A sign-in lasts 30 days; after that the agent asks you to approve it again.

What to do next

Copy the MCP URL from Settings → Connections for the environment you want the outside agent to use. Add it with the Claude Code or Codex commands above, or as a custom OAuth app in ChatGPT with the T3 Connect address. Approve only the sign-in you just started, using a pairing code from Settings → Connections or t3 auth pairing create unless the browser is already signed in as an administrator. Stay on Read only unless you want that agent to start, message and stop threads.