> ## Documentation Index
> Fetch the complete documentation index at: https://docs.sentivel.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Set up with AI

> Point an AI agent at Sentivel and have it set up your workspace.

Sentivel runs a hosted MCP server, so an AI agent (Claude, Cursor, and other
MCP-capable assistants) can read your workspace and make changes on your behalf.
You connect it once, then ask in plain language: "add a monitor for my API and
publish a status page."

## Connect an agent

1. In the dashboard, open **Set up with AI** (on the overview, or during
   onboarding).
2. It mints a personal token and shows the ready MCP configuration for your agent.
3. Paste that into your assistant's MCP settings.

The server lives at `mcp.sentivel.com`. Your agent authenticates as you, scoped to
the workspace you choose, and can only do what your role allows.

<Note>
  A personal token acts as you across your workspaces; pass a workspace with the
  `X-Sentivel-Workspace` header (or let the agent pick one per call). Connected
  apps authorize through OAuth instead and you grant them a specific set of
  workspaces. Revoke either under Account → Connections.
</Note>

## What it can do

The agent has the same reads and actions as the [REST API](/me):
create and manage monitors, components, status pages, incidents, escalation
policies, schedules, and dependencies. People management, billing, and a member's
own phone number stay human-only and are never exposed to a token.

## Example asks

* "Add an HTTP monitor for `https://api.acme.com/health` and attach it to a new
  Checkout component."
* "Publish a status page at `acme` with my components grouped by service."
* "We depend on Stripe for payments. Map that so a Stripe incident shows on our
  page."
* "Open an incident on Checkout and post an update that we're investigating."

Every change an agent makes is a normal workspace change you can see and undo in
the dashboard.