You already write and ship code from Cursor chat. With DevHelm's hosted MCP endpoint, you can create an HTTP monitor from that same chat window with one prompt.
Our post on MCP server monitoring covers keeping MCP servers themselves healthy: tool latency, auth failures, dependency checks. This tutorial is the other direction. You use DevHelm's MCP at mcp.devhelm.io so Cursor can create and manage your monitors.
Why the editor is the right place to create monitors
When you add a /health route, the monitor for that route belongs in the same session. Opening a browser form means copying the URL, picking an interval again, and hoping you remember which regions matter. Monitors deferred that way often stay deferred until production fails.
DevHelm hosts an MCP server so Cursor can call the same product surface you would use in the API or CLI: monitors, incidents, alert channels, and status pages. The dashboard is still the right place for charts and timelines. Create and update work during a coding session can stay in the editor.
If you later want those definitions reviewed in pull requests and deployed from CI, move to monitoring as code. Use this post to get one HTTP monitor live from chat, then decide whether YAML or Terraform should own it.
Prerequisites
- A DevHelm account and API key from Settings → API Keys (keys start with
dh_live_) - Cursor with MCP support enabled
Get posts like this in your inbox
Engineering insights on monitoring, incidents, and uptime. Bi-weekly, 2-minute read.
Connect DevHelm MCP to Cursor in 30 seconds
Add this to .cursor/mcp.json in your project root, or to ~/.cursor/mcp.json for every project:
{ "mcpServers": { "devhelm": { "url": "https://mcp.devhelm.io/mcp", "headers": { "Authorization": "Bearer dh_live_xxxxxxxx" } } } }
Replace dh_live_xxxxxxxx with your real key. Prefer the header form so the token is not embedded in the URL. Restart Cursor, or reload MCP servers, until the DevHelm tools appear in the MCP panel. If tools do not show up, confirm the key still starts with dh_live_ and that the URL is exactly https://mcp.devhelm.io/mcp.
This matches the Cursor config in the MCP overview. The same docs cover Claude Desktop, Windsurf, and the optional local install from PyPI (devhelm-mcp-server) when you want stdio rather than the hosted endpoint.
Create an HTTP monitor with one prompt
In Cursor chat, ask (swap the URL for yours):
Create an HTTP monitor for api.example.com/health checking every 60 seconds from us-east
Cursor should call DevHelm over MCP and create the check. Confirm the reply includes the URL, a 60-second interval, and the us-east region before you move on.
That prompt matches the public MCP examples: HTTP type, health URL, frequency, region. Keep the first request narrow so you can verify the result. Once it works, ask for multi-region coverage or a different interval in a follow-up.
Confirm it in chat (and when to open the dashboard)
Still in the same chat:
List all my monitors and their current status
You should see the new HTTP monitor in the list. If the URL or frequency is wrong, ask Cursor to update it. If the endpoint is not ready yet, ask to pause the check until you finish the route.
Open the dashboard when you want history charts, incident timelines, or status-page layout. Create, list, and update during a coding session can stay in chat.
Optional: install the skill bundle for deeper workflows
MCP gives Cursor live tools against your workspace. For longer configure and investigate flows that also understand the CLI and Monitoring-as-Code layouts, install the skill bundle:
npx devhelm skills install --target cursor
Install steps, auth via DEVHELM_API_TOKEN, and the skills-vs-MCP comparison are in the AI skills overview. Run both if you want: MCP for in-session tool calls, skills for guided workflows and repo-aware edits.
What to do next
- Read the MCP overview for Claude Desktop, Windsurf, and local stdio setup
- When you want definitions in git, follow monitoring as code
- If you run your own MCP servers, see MCP server monitoring
Try it
Grab a free API key at app.devhelm.io, add https://mcp.devhelm.io/mcp to Cursor, then ask: Create an HTTP monitor for my API health endpoint.
Get posts like this in your inbox
Engineering insights on monitoring, incidents, and uptime. Bi-weekly, 2-minute read.