Connect Codex

Codex can connect directly to a remote Streamable HTTP MCP server. The ChatGPT desktop app, Codex CLI, and Codex IDE extension share MCP configuration on the same machine. Once connected, Codex can build and review seating charts and set up events in SeatLayer. It cannot hold, book, or sell seats; see what the Designer MCP covers. OpenAI documents its configuration in its MCP documentation.

Desktop app or IDE

1. Open Settings → MCP servers.
2. Select Add server.
3. Name it SeatLayer, choose Streamable HTTP, and paste the URL.
4. Save and restart the app or extension.
5. Select Authenticate. SeatLayer opens in your browser.
6. Approve the grant on the SeatLayer consent screen (below).
7. Use /mcp to confirm that the server is connected.

CLI

Add the remote server to ~/.codex/config.toml or the trusted project’s .codex/config.toml:

[mcp_servers.seatlayer]
url = "https://mcp.seatlayer.io/mcp"
auth = "oauth"
default_tools_approval_mode = "writes"

Then run codex mcp login seatlayer to authenticate and codex mcp list to check it. default_tools_approval_mode = "writes" keeps Codex asking before any tool that changes a chart. OpenAI documents the current configuration keys and CLI commands in its MCP guide. Do not place a SeatLayer secret API key in a prompt or configuration file.

The SeatLayer consent screen

The person approving needs the SeatLayer admin role. They choose:

• the Workspace the agent works in;
• One chart (default) or Every chart in this workspace; and
• whether to allow Manage live events, which lets the agent close sections and block seats on events already on sale. It is off by default.

Access tokens expire hourly and refresh on their own.

Verify safely

Start with a read and capability check:

Use the SeatLayer connector.
Call get_capabilities first, state the authorized chart or workspace, and
summarize the current chart. Do not change, approve, or publish anything.

The reply should name the chart or workspace you approved and list the tools the grant allows. Then follow the staged workflow in Designer MCP. Any write must use the current expected revision, and publishing still requires a human-approved unchanged revision.