> ## Documentation Index
> Fetch the complete documentation index at: https://docs.devhelm.io/llms.txt
> Use this file to discover all available pages before exploring further.
# MCP Server
> Give your AI coding assistant access to DevHelm monitors, incidents, alerting, status pages, and more — in one line of config
The DevHelm MCP server gives AI coding assistants like Cursor, Claude Desktop, and Windsurf direct access to your monitoring stack. Create monitors, investigate incidents, manage alert channels, and configure status pages — all through natural language in your editor.
DevHelm hosts the server at `mcp.devhelm.io` so there's nothing to install. Add the URL to your MCP client config and you're connected.
## Connect in 30 seconds
All you need is your API token and one config entry. Grab a token from **Settings → API Keys** in the [Dashboard](https://app.devhelm.io).
Add to `.cursor/mcp.json` in your project root (or `~/.cursor/mcp.json` for global access):
```json theme={null}
{
"mcpServers": {
"devhelm": {
"url": "https://mcp.devhelm.io/mcp",
"headers": {
"Authorization": "Bearer dh_live_xxxxxxxx"
}
}
}
}
```
Add to your Claude Desktop config:
* **macOS:** `~/Library/Application Support/Claude/claude_desktop_config.json`
* **Windows:** `%APPDATA%\Claude\claude_desktop_config.json`
```json theme={null}
{
"mcpServers": {
"devhelm": {
"url": "https://mcp.devhelm.io/mcp",
"headers": {
"Authorization": "Bearer dh_live_xxxxxxxx"
}
}
}
}
```
Add to your Windsurf MCP settings:
```json theme={null}
{
"mcpServers": {
"devhelm": {
"url": "https://mcp.devhelm.io/mcp",
"headers": {
"Authorization": "Bearer dh_live_xxxxxxxx"
}
}
}
}
```
Any MCP-compatible client that supports Streamable HTTP can connect to the hosted server.
**If your client supports custom headers** (recommended):
```
URL: https://mcp.devhelm.io/mcp
Header: Authorization: Bearer dh_live_xxxxxxxx
```
**If your client only accepts a URL** (token in path):
```
URL: https://mcp.devhelm.io/dh_live_xxxxxxxx/mcp
```
The path-based form embeds your token in the URL. Treat the entire URL as a secret — don't commit it to a public repo. Prefer the header form whenever possible.
That's it. Restart your MCP client and the DevHelm tools will appear.
## What you can do
Once connected, talk to your assistant like you would to a teammate:
* **"List all my monitors and their current status"** — get a quick overview without leaving your editor
* **"Create an HTTP monitor for api.example.com/health checking every 60 seconds from us-east"** — spin up monitoring in seconds
* **"Show me active incidents and the recent check results for the failing one"** — investigate without context-switching to the dashboard
* **"Set up a Slack alert channel with this webhook URL"** — wire up notifications conversationally
* **"Add api.example.com as a component on the public status page"** — manage your status page from chat
* **"What's the uptime for the payments service over the last 30 days?"** — pull reporting data inline
The MCP server covers the entire DevHelm product surface — anything you can do in the dashboard or CLI, your AI assistant can do too.
## Available tools
The server exposes 120 tools across 17 categories:
| Category | Tools |
| ------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Monitors** | list, get, create, update, delete, pause, resume, test, results, versions |
| **Incidents** | list, get, create, resolve |
| **Alert channels** | list, get, create, update, delete, test |
| **Notification policies** | list, get, create, update, delete, test |
| **Resource groups** | list, get, create, update, delete, add/remove members |
| **Tags** | list, get, create, update, delete |
| **Environments** | list, get, create, update, delete |
| **Secrets** | list, create, update, delete |
| **API keys** | list, create, revoke, delete |
| **Webhooks** | list, get, create, update, delete, test |
| **Status pages** | list, get, create, update, delete, components, groups, incidents, subscribers, domains |
| **Forensics** | incident timeline, check trace, policy snapshot, rule evaluations, transitions |
| **Services** | search, get, live status, summary, categories, components, uptime, incidents, incident detail, day rollup, component uptime, all components uptime, maintenances |
| **Dependencies** | list, get, track, update alert sensitivity, delete |
| **Maintenance windows** | list, get, create, update, cancel |
| **Deploy lock** | acquire, current, release, force-release |
| **Status / health** | overview |
See the [Tools reference](/sdk/mcp/tools-reference) for the full per-tool catalog with parameters.
## Alternative: run locally
If you prefer to run the server on your own machine (offline use, air-gapped environments, or development), the same code is published to PyPI as `devhelm-mcp-server`. It speaks MCP over stdio.
```bash theme={null}
uvx devhelm-mcp-server
```
See [Configuration → Run locally](/sdk/mcp/configuration#run-locally) for full local setup instructions and per-client config snippets.
## Next steps
Authentication modes, environment variables, local setup, and troubleshooting.
Full per-tool catalog with parameters and return shapes.