Skip to main content

TrackingTime MCP Server for AI Assistants

Learn how to connect AI assistants like Claude, ChatGPT, Cursor, and more to your TrackingTime workspace using the Model Context Protocol (MCP).

Eric Tulle avatar
Written by Eric Tulle
Updated yesterday

Overview

The TrackingTime MCP Server allows you to connect AI assistants and coding agents directly to your TrackingTime workspace. Once connected, you can use natural language to track time, query your projects and tasks, list team members, and much more — all without leaving your AI tool.

MCP (Model Context Protocol) is an open standard that lets AI applications interact with external services. TrackingTime's MCP server is hosted remotely, so there's no installation required. Just connect and start working.

Server URL: https://mcp.trackingtime.co/sse


What You Can Do

Once connected, the following tools are available to your AI assistant:

Tool

Description

Get Profile

Retrieve your user profile, account ID, and workspace info.

Create Time Entry

Log a new time entry with start/end times, timezone, notes, and optional task or project assignment.

Update Time Entry

Modify an existing time entry (change notes, start or end times).

List Time Entries

Query time entries by date range. Filter by user or across the entire team (company-wide). Supports pagination.

Get Time Entry

Retrieve details of a single time entry by its ID.

List Users

View all workspace members. Filter by active, archived, or all. Includes optional billing and custom field data.

List Projects

View all projects in your workspace. Filter by active, archived, or all.

List Tasks

View tasks in your workspace, optionally filtered by project.

List Customers

View all customers/clients in your workspace.

List Services

View all services defined in your workspace.


Requirements

Before connecting, make sure you have the following:

  1. A TrackingTime account (sign up here if you don't have one).

  2. Your TrackingTime username (the email you use to log in).

  3. An App Password generated from your TrackingTime workspace settings.

  4. An AI tool that supports MCP, such as Claude Desktop, Claude Code, Cursor, Windsurf, VS Code, ChatGPT, or any MCP-compatible client.


Generate an App Password

The MCP server uses Basic Authentication with your username and an app password. To generate one:

  1. Log into TrackingTime.

  2. Go to SettingsApp Passwords (or IntegrationsAPI).

  3. Create a new App Password.

  4. Copy the generated password — you'll need it for the configuration step.

⚠️ Keep your app password secure. Do not share it publicly. You can revoke it at any time from your settings.


Connect with Claude Desktop

To connect TrackingTime with Claude Desktop:

  1. Open Claude Desktop.

  2. Go to SettingsDeveloperEdit Config.

  3. Add the following to your claude_desktop_config.json file:

{
"mcpServers": {
"trackingtime": {
"command": "npx",
"args": [
"mcp-remote",
"https://mcp.trackingtime.co/sse"
]
}
}
}
  1. Save the file and restart Claude Desktop.

  2. You should see the TrackingTime tools available (look for the hammer 🔨 icon).


Connect with Cursor

To connect TrackingTime with Cursor:

  1. Open Cursor SettingsMCP.

  2. Click Add new global MCP server.

  3. Paste the following configuration:

{
"mcpServers": {
"trackingtime": {
"command": "npx",
"args": [
"mcp-remote",
"https://mcp.trackingtime.co/sse"
]
}
}
}
  1. Save and restart Cursor.


Connect with Windsurf

To connect TrackingTime with Windsurf:

  1. Open the MCP section in the right panel.

  2. Edit your mcp_config.json file and add:

{
"mcpServers": {
"trackingtime": {
"command": "npx",
"args": [
"mcp-remote",
"https://mcp.trackingtime.co/sse"
]
}
}
}
  1. Click Refresh in the MCP panel.

  2. TrackingTime should appear under "available MCP servers".


Connect with Claude Code (CLI)

If you use Claude Code from the terminal:

claude mcp add trackingtime -- npx mcp-remote https://mcp.trackingtime.co/sse

That's it! The server will be available in your next Claude Code session.


Connect with ChatGPT

To use TrackingTime as a connector in ChatGPT:

  1. Go to ChatGPT SettingsConnectors.

  2. Add a new custom MCP server.

  3. Enter the URL: https://mcp.trackingtime.co/sse

  4. Follow the authentication steps.

Note: ChatGPT MCP support requires Developer Mode (available for Pro/Plus users under Settings → Connectors → Advanced → Developer mode).


Connect with Any Other MCP Client

For any MCP-compatible client that supports SSE (Server-Sent Events) transport, use the following endpoint:

URL: https://mcp.trackingtime.co/sse

For clients that require the mcp-remote bridge (stdio-based clients):

{
"mcpServers": {
"trackingtime": {
"command": "npx",
"args": [
"mcp-remote",
"https://mcp.trackingtime.co/sse"
]
}
}
}

Usage Examples

Once connected, you can interact with TrackingTime using natural language. Here are some examples:

Tracking time:

  • "Log 2 hours today on the Website Redesign project from 9am to 11am."

  • "Create a time entry for yesterday on the 'Bug Fixes' task with the note 'Fixed login issue'."

Querying your data:

  • "Show me all my time entries from last week."

  • "How much time did the team track this month?" (uses company-wide filter)

  • "List all active projects in my workspace."

Managing your workspace:

  • "Show me all team members in my workspace."

  • "List all tasks in the Mobile App project."

  • "What customers do we have in TrackingTime?"

💡 Tip: When asking for team-wide data (e.g. "who worked the most this week?"), the AI will automatically use the company-wide filter to get all data in a single request.


Authentication

Every request requires your username and app password. Depending on the MCP client, you may be prompted for these credentials:

  • On the first interaction with any TrackingTime tool.

  • Stored for the duration of your session (not persisted between restarts in most clients).

The MCP server uses Basic Authentication over HTTPS. Your workspace (account_id) is auto-detected from your credentials — you don't need to provide it manually.


Common Limitations and FAQs

  • The MCP server is read/write for time entries — you can create and update events. Projects, tasks, customers, services, and users are read-only (list/query).

  • App passwords are required — your regular login password won't work. You must generate an App Password from your TrackingTime settings.

  • Node.js is required for stdio clients — if using the npx mcp-remote bridge, you need Node.js 18+ installed on your machine.

  • The SSE endpoint requires no local installation — the server is fully hosted by TrackingTime.

  • Pagination — list endpoints support page and page_size parameters. Your AI assistant will handle pagination automatically when querying large datasets.


Supported Clients

Client

Transport

Status

Claude Desktop

stdio (via mcp-remote)

✅ Supported

Claude Code

stdio (via mcp-remote)

✅ Supported

Cursor

stdio (via mcp-remote)

✅ Supported

Windsurf

stdio (via mcp-remote)

✅ Supported

VS Code (Copilot)

stdio (via mcp-remote)

✅ Supported

ChatGPT

SSE (remote)

✅ Supported

Cline (VS Code)

stdio (via mcp-remote)

✅ Supported

Any MCP Client

SSE

✅ Supported


Need Help?

If you have questions or run into issues with the MCP server, feel free to reach out:

Did this answer your question?