New · Remote MCP server

Give your AI tools a safe way to operate monitoring.

Monitoring.app for MCP gives compatible AI clients scoped access to investigate monitors, manage safe settings, and prepare status communication—with explicit controls around writes and publishing.

Monitoring.app MCP
Remote
Streamable HTTP endpoint
https://api.monitoring.app/mcp
Which monitors failed in the last hour?
I found two failed checks. The journey definition and request secrets remain private.
Draft a customer update. Do not publish it.
Draft saved privately. Publishing remains a separate action.
24
Tools
6
OAuth scopes
24h
Safe retries
Current MCP capabilities

Read the signal. Prepare the response. Act with control.

The first MCP release focuses on the work that benefits most from context: monitor operations and customer-facing status communication.

1 tool

Workspaces

W

Start in the right customer or operational context.

  • List the active workspaces linked to the signed-in account.
6 tools

Monitors

M

Investigate health and manage safe monitor settings.

  • List HTTP, SSL, and journey monitors.
  • Get current scheduling state and bounded recent results.
  • Create HTTP or SSL monitors with plan-valid intervals and regions.
  • Update names, targets, alert thresholds, regions, and enabled state.
  • Delete a monitor only after exact-name confirmation.
9 tools

Status pages

S

Prepare and maintain customer-facing service communication.

  • List drafts and published status pages, then inspect their components.
  • Create private drafts and update names, slugs, descriptions, and timeline settings.
  • Publish or unpublish with a separate, explicit action.
  • Add, rename, reorder, and remove monitor-backed components.
  • Delete only after exact-name confirmation; attached custom domains block deletion.
8 tools

Incidents

I

Draft, publish, and resolve updates without losing the audit trail.

  • List and inspect private or published incidents for a status page.
  • Create an incident draft with category, state, timing, and affected components.
  • Update private metadata and add private update drafts.
  • Publish one selected update with an explicit action.
  • Draft and publish postmortems as separate steps.
Safe by design

Autonomy with boundaries you can explain.

MCP requests pass through the same workspace permissions and product rules as the portal, with extra safeguards for machine-driven operations.

1

OAuth-protected access

Remote clients authenticate with linked accounts and request only the scopes their tools need.

2

Workspace boundaries

Every operation names a workspace. Membership and role checks are enforced again server-side.

3

Safe retries

Mutations require idempotency keys, with operation receipts retained for 24 hours.

4

Explicit publishing

Drafting and publishing are separate actions for status pages, incident updates, and postmortems.

5

Guarded deletion

Destructive tools require an exact current name and are marked destructive in the MCP catalog.

6

Secret-safe reads

Journey headers, bodies, URLs, definitions, and request secrets stay out of MCP responses.

Ask naturally

Bring operational context into the conversation.

Your AI client decides how to present and combine tools. Monitoring.app supplies the typed operations, permissions, and structured results underneath.

Show failing monitors in my customer operations workspace.
Create an SSL monitor for api.example.com every five minutes.
Draft a status-page incident for the affected API components.
Publish the approved incident update, then show me the result.
Connect in three steps

A remote server, not another agent to maintain.

01

Add the endpoint

Use the Monitoring.app MCP URL in a compatible client that supports remote Streamable HTTP.

02

Approve access

Sign in with your linked account and approve only the workspace, monitor, and status-page scopes you need.

03

Choose a workspace

Start with a read request, then allow deliberate writes or publication actions when the workflow calls for them.

Current boundary: HTTP journey creation and editing remain portal-only. MCP can return safe journey metadata and bounded results, but never journey definitions, request headers, bodies, or private targets. Workspace members can read; administrative writes require an admin or owner role.

Ready to put monitoring in the loop?

Connect a compatible MCP client to investigate faster, prepare clearer updates, and keep every action inside your workspace boundaries.