MCP Server Setup
BeatOS includes a Model Context Protocol (MCP) server that lets AI assistants like Claude, ChatGPT, and other MCP-capable clients manage your tasks, habits, reviews, and more through natural conversation.
Prerequisites
- A BeatOS account
- An API key (see Authentication)
Quick Setup (Remote — Recommended)
Connect any MCP-capable client directly to BeatOS over HTTPS. No local install needed.
Claude Code / Claude Desktop
Add to your .mcp.json (project-level) or Claude Desktop config:
{
"mcpServers": {
"beatos": {
"type": "http",
"url": "https://mcp.beatos.app/api/mcp",
"headers": {
"Authorization": "Bearer pgtd_your_api_key_here"
}
}
}
}
ChatGPT and other MCP clients
Use these connection details in your client's MCP settings:
| Field | Value |
|---|---|
| Server URL | https://mcp.beatos.app/api/mcp |
| Transport | Streamable HTTP |
| Auth Header | Authorization: Bearer pgtd_your_api_key_here |
Replace pgtd_your_api_key_here with your API key from Settings → API Keys. (API keys retain the pgtd_ prefix for backwards compatibility with existing keys.)
Zero Install
The remote MCP endpoint runs on BeatOS's servers — no local build, no Node.js, no cloning. Just add the URL and your API key.
Alternative: Local stdio via npm
For offline work, custom proxying, or clients that don't support HTTP MCP, run the server locally as an npm package. No repo clone required.
# Verify the package is installed and reachable
npx @beatos/mcp-server --help
npx will download and cache the package on first run. Point your MCP client at it:
{
"mcpServers": {
"beatos": {
"command": "npx",
"args": ["@beatos/mcp-server"],
"env": {
"API_BASE_URL": "https://mcp.beatos.app",
"API_TOKEN": "pgtd_your_api_key_here"
}
}
}
}
To pin a specific version (recommended for stability):
"args": ["@beatos/mcp-server@0.1.0"]
Available Tools
The MCP server exposes 18 tools:
Task Management
| Tool | Description |
|---|---|
list_tasks | List tasks with optional date and completion filters |
create_task | Create a new task with title, date, urgency, importance |
update_task | Update task fields |
complete_task | Mark a task as completed |
delete_task | Delete a task |
Projects
| Tool | Description |
|---|---|
list_projects | List all projects |
create_project | Create a new project |
Habits
| Tool | Description |
|---|---|
list_habits | List all habits |
create_habit | Create a new habit |
log_habit | Log a habit completion or skip |
get_habit_streaks | Get streak data for all habits |
Reviews & Analytics
| Tool | Description |
|---|---|
list_reviews | List review entries |
create_review | Create a weekly or monthly review |
get_dashboard | Get dashboard statistics |
Organization
| Tool | Description |
|---|---|
list_buckets | List life buckets |
list_bottlenecks | List bottlenecks |
create_bottleneck | Create a bottleneck |
Usage Examples
Once configured, you can interact with BeatOS through your AI assistant:
"Create a task called 'Review Q1 metrics' for tomorrow, marked as Important"
"What habits do I have due today?"
"Run my weekly review"
"Show me my dashboard stats"
"Log my morning meditation habit as complete"
Troubleshooting
API Key Security
Never commit your API key to version control. Use environment variables or a .env file that's in your .gitignore.
- "Unauthorized" errors: Verify your API key is valid in Settings → API Keys
- Tools not appearing: Restart your AI client after adding/modifying MCP config
- Timeout errors: The remote endpoint has a cold-start time of ~2s on first request
- Local
npxerrors: Runnpx clear-npx-cacheornpm cache clean --force, then retry. Make sure Node 18+ is installed (node --version).