Claude + MCP
Ask Claude if anything is down
Connect Sentinel to Claude over MCP and your monitoring becomes a conversation: live status, incident triage, certificate deadlines, and a global URL probe, all by asking.
Sentinel → Claude
Why connect Claude
Monitoring that answers questions
A dashboard tells you everything, which is another way of saying it makes you find the answer yourself. Connecting Sentinel to Claude over MCP turns monitoring into something you ask: whether anything is down, which certificates expire this month, what happened to a client site overnight, and whether that slowdown is regional.
It also puts monitoring inside the loop where work already happens. Debugging a deploy with Claude Code? The same conversation can pause the monitor first and confirm the site is healthy after, without a tab switch.
- Works with Claude Desktop, Claude Code and claude.ai connectors
- Read tools for status, incidents, history and certificates
- Write tools on Pro and Business: create, update, pause, resolve
- Published in the official MCP registry as io.rootstuff/sentinel
Is anything down right now?
One incident: acmestudio.co, down 4 minutes. Connection timeout from Ashburn.
One thing worth knowing: clientbakery.com's certificate expires in 14 days.
We're redeploying it now, pause the monitor
Paused acmestudio.co. Say the word when you want checks running again.
Features
What Claude can do with Sentinel
Ask, do not dig
'Is anything down right now?' and 'what needs my attention?' are single questions to Claude, answered from your live monitors instead of a dashboard you have to open.
Incident triage in conversation
List open incidents, read what failed and from which regions, acknowledge and resolve, all without leaving the conversation you were already working in.
Probe any URL globally
The check_url tool tests any address from Sentinel's regions on demand, so 'is it down for everyone or just me' has an authoritative answer inside Claude.
Write access when you want it
On Pro and Business, Claude can create monitors, pause them during a deploy, and update incident state. Read-only tokens keep it to questions.
In the official MCP registry
Sentinel is published as io.rootstuff/sentinel in the MCP registry, so clients that browse the registry can find and connect it without hand-typed config.
Scoped, revocable tokens
Access runs through API tokens with explicit create, update and delete permissions. Give Claude exactly the reach you are comfortable with, and revoke it any time.
How do I connect Claude to Sentinel?
Create an API token in Sentinel, then add the Sentinel MCP server to Claude Desktop or Claude Code with that token. On claude.ai, add it as a connector and authorise with OAuth. The setup guide walks through each client.
What can Claude actually do with my monitors?
With a read token: report what is up and down, summarise incidents, list expiring certificates, pull response time and DNS history, and probe any URL from Sentinel's regions. With full API access on Pro or Business, it can also create and update monitors, pause them, and acknowledge or resolve incidents.
Which plans include this?
API access starts on the Starter plan with read-only tools. Write tools require full API access, which is included on Pro and Business.
Is it safe to give an AI access to my monitoring?
Access is scoped by the token you issue. A read-only token cannot change anything, write actions require explicit token permissions, and deletes are refused without the delete permission. Revoking the token cuts access instantly.
Does this work with AI tools other than Claude?
Sentinel speaks standard MCP, so any client that implements the protocol can connect. Claude Desktop, Claude Code and claude.ai are the clients we document and test against.
Put your monitoring on speaking terms
Create a token, connect the MCP server, and ask your first question in a few minutes.
No card for the free plan · 14-day trial on paid plans · Cancel anytime