When something breaks, the first question is always the same: what is down, and since when? Until now the answer lived in the Site Qwality dashboard. Today it also lives wherever you already work with an AI assistant.

Site Qwality now runs a hosted Model Context Protocol (MCP) server. Add it to Claude, ChatGPT, Cursor, VS Code or any MCP client, sign in with your Site Qwality account, and your assistant can read your monitors, incidents, status pages, logs, metrics and traces, and take the everyday actions you allow. It is included on every plan, including the free one.

The server lives at:

https://mcp.siteqwality.com/mcp

What you can ask

The assistant answers from your own account, not from general knowledge. A few things people ask on day one:

  • "What needs attention right now?" It lists monitors that are down or degraded and unresolved incidents, most urgent first.
  • "Why is the checkout API failing?" It reads the monitor's recent checks: status codes, response times, the error, and for browser checks, the step that failed.
  • "What did the logs say around 14:00?" It searches your logs by text, level and source over the window you name, and can pull the metrics and traces from the same minutes.
  • "Summarize last night's incident." It reads the incident with its timeline, responders and affected monitors.

And, when you allow it, the things you would otherwise click through:

  • Pause a monitor for a deploy, and resume it afterwards.
  • Acknowledge an incident, post updates (on its status page too, if it is published), and resolve it.
  • Create or edit an HTTP monitor.

Twenty-three tools cover those. Two more find and call any other documented operation in the Site Qwality API, so the assistant is not limited to the ones we curated.

You decide what it can touch

There are no API keys to paste. When you add the server, your assistant sends you to Site Qwality to sign in, and you see exactly which app is asking and what it wants. Access is granted area by area (monitors, incidents, notifications, status pages, logs, metrics, traces, real user monitoring and dashboards), each as read or read and write. A single switch makes the whole connection read-only.

Three rules hold however you set it up:

  • It acts as you, never above you. The assistant has your role and nothing more. If you are a viewer, your assistant is read-only too.
  • Some things are out of reach. Billing, team members, API keys and SSO settings cannot be reached through MCP at all, whatever you approve.
  • Deletes are marked. Tools that delete are flagged as destructive, so assistants that honour the flag ask you before running them.

Every change is on the record

An assistant that can change your monitoring should leave a trail you can follow. Every change one makes shows up in your audit log as the person who approved it, with the assistant's name, the tool it used, and a mark that it came through MCP. Connecting an assistant, refusing one, and disconnecting one are logged too, with the access that was granted.

If you change your mind, disconnect it under AI Assistants in your account settings. It cannot sign in again without you, and the access it already had stops working within a minute.

Setting it up

The setup guide has steps for Claude, Claude Code, ChatGPT, Cursor, VS Code and the Codex CLI. In most of them it is one line or one form: add the server URL, then sign in when asked. In Claude Code, for example:

claude mcp add --transport http siteqwality https://mcp.siteqwality.com/mcp

Agents that cannot open a browser, such as a CI job or a scheduled script, can use a scoped API key instead, sent as a Bearer token. The key's scopes apply to every tool exactly as they do on the REST API, and its changes are audited as that key, marked as coming through MCP.

Why a hosted server

Most MCP servers still run on your own machine: you install a package, keep it updated, and paste a long-lived key into a config file. We host ours instead. There is nothing to install or upgrade, sign-in uses your normal Site Qwality login, and every tool call goes through the same API and permissions as the dashboard, with every change in the same audit log. When we add a feature to the API, your assistant can use it the same day.

It costs nothing extra on any plan. Plan limits and usage quotas apply exactly as they do in the dashboard.

Read the setup guide, or start free and connect your assistant in a couple of minutes.