Pulse runs a remote MCP (Model Context Protocol) server, so AI coding tools like Claude Code, Codex, and Gemini CLI can read your timesheet and log time entries for you — no manual stopwatch. Once connected, you can ask your agent things like "how much time have I logged this week?" or have it log an entry automatically after finishing a task.
This guide walks through connecting from each tool. The setup is the same everywhere: point the tool at the Pulse MCP endpoint and give it a credential.
| Endpoint | https://pulse.transmit.work/api/mcp |
| Transport | Streamable HTTP (remote) |
| Auth | Authorization: Bearer <token> — either a Pulse API token (psk_…) or OAuth 2.1 |
👥 Working with multiple clients? See Users & roles → multi-client pattern for the folder-scoped setup.
🔒 What Pulse stores, who can see it, the data-flow diagrams? See Privacy & data isolation.
You need a Pulse account and a credential. There are two options:
psk_…) — a long-lived bearer token. Best for CLI tools (Claude Code, Codex, Gemini).⚠️ Token creation is Admin-only. The in-app Settings → API & Connections screen (where tokens and OAuth clients are created) is visible to workspace Admins only. If you're a Member, ask an Admin to create a token for you, or connect via OAuth through a client an Admin has registered.
If you're an Admin:
claude-code-laptop) and create the token.psk_… value — it's shown only once. Store it somewhere safe.If you're a Member: ask an Admin to create a token for you. The token is tied to your user — entries logged with it land on your timesheet, not the Admin's.
https://claude.ai/api/mcp/auth_callback.http://localhost:PORT/callback URI (any port). You'll pass the same port as --callback-port when adding the server.