If an AI agent helps run your agency, it should be able to answer “is anything broken across our client sites?” without you alt-tabbing to a dashboard.
Formitor now speaks MCP (Model Context Protocol), the open standard that Claude and a growing list of agents use to talk to tools.
What the agent can do
- List your sites and their current health
- Ask what is broken right now across the whole portfolio
- Pull a single site’s full status: forms, checks, uptime, deliverability
- Read your recent alerts
So mid-conversation, your agent can answer things like “did anything fail on the Henderson site this week?” or “summarize monitoring status before my Monday client calls” with live data.
Setup
From More → AI access in the dashboard, generate a token and add it to your agent’s MCP configuration. Tokens are shown once, can be revoked anytime, and each token is scoped to your account.
The safety model
Access is read-only. An agent can see monitoring state; it cannot pause monitoring, change settings, delete sites, or touch anything. If a token leaks, the blast radius is “someone can read your monitoring dashboard”, and you revoke it in one click.
That’s deliberate. Monitoring data is exactly the kind of context agents are good with: summarizing, correlating, flagging what needs a human. Mutating your monitoring config is not a job to hand to an autocomplete, so the API simply doesn’t allow it.
Why we shipped it
Agencies are already gluing AI into their ops. The tools that win that world are the ones agents can read natively instead of screen-scraping. Your monitoring just became one of them.