Use Pulse with OpenAI products through its remote MCP server. The setup differs between local Codex clients and ChatGPT web Work mode, so choose the section matching where you work.

🔑 You need a Pulse account and credential first. See Connect Pulse to your AI tools (MCP) for token creation, roles, verification, prompt examples, and the complete tool reference.

Codex CLI

API tokens are the simplest setup. Keep the token in an environment variable rather than writing the raw value into config.toml.

export PULSE_TOKEN="psk_your_token_here"

codex mcp add pulse \
  --url <https://pulse.transmit.work/api/mcp> \
  --bearer-token-env-var PULSE_TOKEN

Verify the connection:

codex mcp list

Inside the Codex terminal interface, run /mcp to see active servers. Ask Codex to call pulse_whoami to confirm that the credential maps to the correct Pulse account.

The environment variable must be available to the process that launches Codex. A variable exported in one terminal may not be available to a desktop app launched from Finder, Dock, or another shell.

Configure manually

Codex stores personal MCP configuration in ~/.codex/config.toml. A trusted project can instead use .codex/config.toml.

[mcp_servers.pulse]
url = "<https://pulse.transmit.work/api/mcp>"
bearer_token_env_var = "PULSE_TOKEN"

Streamable HTTP MCP is supported directly; no experimental MCP client flag or mcp-remote bridge is required.

ChatGPT desktop app / Codex

The desktop app, Codex CLI, and Codex IDE extension share MCP configuration on the same Codex host.

1. Open Settings → MCP servers.
2. Select Add server.
3. Enter the name pulse.
4. Choose Streamable HTTP.