Developer surface

PulseTrace + MCP server

Let Claude, Cursor and your own agents run your monitoring.

PulseTrace ships a Model Context Protocol server, so an AI assistant can list monitors, add one, read an incident timeline or pull a report — governed by the same OAuth grants, scopes and org policy as any human user.

Set it up in three steps

1Settings → Developer & API → MCP: enable MCP for the organization and choose the access level (read-only, monitoring operator, full management, or a custom scope set).
2Register the client (Claude Desktop, Claude Code, Cursor or your own) and complete the OAuth consent — every grant is listed and revocable under Connected Apps.
3Ask: "Add an HTTP monitor for api.acme.io from three regions and alert #incidents." Tool calls appear in the MCP activity log with the acting user.

Start monitoring free Read the docs

Prompt

> Add an HTTPS monitor for https://api.acme.io/health,
  check every 30s from us-east, eu-west and ap-south,
  alert the #incidents Slack destination on 2 failed checks.

✓ Created monitor "api.acme.io health" (id 8f3a…)
✓ Alert rule: down after 2 failures → Slack #incidents

Questions about MCP server

Can an agent do anything a user can?

Only what its grant allows. Access levels and custom scopes are set per organization, tool calls are logged with the granting user, and any grant can be revoked at any time.

Which clients are supported?

Any MCP-compatible client over OAuth — Claude Desktop, Claude Code, Cursor and custom agents have been used in testing.

More developer surfaces

REST APIEverything the dashboard does, as JSON.OpenAPI importTurn a spec into a monitor per endpoint.Probe agentCheck from inside your own network.