Reconnecting
Restoring connection…
Reconnecting
Restoring connection…
Docs navigation
Get started
AI agents
Connect
The fleet
Operate
Day to day
When it breaks
Govern access
Team & account
Access
Identity concepts
Provider guides
Account
ChatGPT
Connect ChatGPT to emisar through Developer mode over remote MCP. OAuth signs you in — you do not need to copy an agent key.
Using Claude Code, Cursor, Claude Desktop, or a CLI instead? Those connect through the
emisar-mcp
stdio bridge — see Connect a CLI agent. For Claude.ai, see Claude.ai.
- An account with at least one runner online. Otherwise the agent signs in fine and finds an empty catalog with nothing to call. New here? Run the quickstart first, then come back to wire in your LLM.
- Runners that do not require signed dispatch. Claude.ai and ChatGPT do not create emisar dispatch signatures, so a runner that requires one needs the local MCP bridge with a signing key and certificate instead.
Add the connector#
ChatGPT reaches a custom MCP server through Developer mode. Access depends on your account and workspace policy.
- Open Settings → Security and login. Turn on Developer mode.
- Open ChatGPT Plugins and select the small + button just right of the search box — it is easy to miss.
-
Fill the form:
- Name — paste the name below
- Description (optional) — "Run approved infrastructure actions on my fleet."
- Connection → Server URL — paste the URL below
- Authentication — choose OAuth. There is no client ID or secret to paste; emisar registers the connector itself.
Name
emisarRemote MCP server URL
https://emisar.dev/api/mcp/rpcSelect Create, then review the discovered tools and metadata.
- ChatGPT opens the emisar sign-in page. Sign in and choose Authorize. ChatGPT then lists the emisar tools.
- Review the connection permissions. emisar policy, approvals, and audit still apply to every call.
- Start a new chat. Add emisar from the tools menu. A listed tool means metadata access only. Adding the connection lets the chat call those tools.
- After a tool metadata change, open the plugin connection. Refresh its tool metadata before the next test.
Verify the connection#
-
Ask the client to run
linux.uptimeon any one of the connected runners. - Confirm that the client returns the action output.
- Open Audit. Confirm that the event records the client, runner, action, and reason.
What the sign-in grants#
The consent screen mints a key bound to your membership. The connector acts as you. Every call is attributed to you and the agent you are using. It reaches only the runners and actions your own access already allows (how a key inherits its operator's scope). Revoke it any time from AI agents in the console, or from the connector settings in Claude or ChatGPT.
Point your agent at real infrastructure — safely.
Set up a runner first, then connect ChatGPT to your account's MCP endpoint and let policy decide what runs.
Three runners. Seven-day audit. No credit card.