Ask Your Readiness Data Questions in Plain English: Connect Civasure to Claude with MCP

Most readiness questions don’t start as a database query. They start as a sentence: “How many households have we reached this month, and which ZIP codes are still lagging?” Getting from that sentence to an answer usually means opening a dashboard, picking filters, and reading a chart.

Civasure now offers a shortcut. We run a remote MCP server — a standard way for AI assistants like Claude to call live tools — so you can ask those questions in plain language and get an answer pulled straight from your data. No SQL, no exports.

This post covers what it does, what it can’t do, and how to connect it. The full setup lives in the developer docs.

What is MCP, briefly

The Model Context Protocol is an open standard that lets AI assistants connect to external tools and data. When you connect Civasure, your readiness data becomes a set of tools the assistant can call on your behalf — so instead of describing your numbers to it, it reads them itself, on demand.

Concretely, the Civasure MCP server exposes three read-only tools:

• Stats — submission totals, households reached, ZIP codes with data, and the most recent submission time for your area.
• Coverage — per-ZIP submission and household counts, so you can see where participation is strong and where it’s thin.
• Readiness — aggregate distributions for each readiness survey question (for example, what share of households report an evacuation plan).

Ask “Which ZIPs are lagging on participation?” or “What’s our evacuation-plan rate, and where is it lowest?” and the assistant calls the right tool and answers from the live numbers.

Aggregate, anonymous, and scoped — by design

The same rules that govern the Civasure dashboard govern the MCP server. It only ever returns counts and distributions — never an individual response, never a device identifier, never anything that could identify a person. And a connection can only read aggregates inside your organization’s geographic area — the exact area your dashboard already shows.

In other words, connecting an AI assistant doesn’t widen what’s exposed. It’s a new way to ask the same anonymous, scoped questions.

Connecting it

API access — including MCP — is available on the Regional and Government plans. The setup takes a couple of minutes for someone comfortable editing a config file.

1. Create an API key. In the dashboard, go to Settings → API and create a key. You’ll see it once, so copy it somewhere safe.

2. Add the server to your client. Desktop assistants run MCP servers as a local command, so a remote server like ours is reached through the mcp-remote bridge (Node.js required). In Claude Desktop’s config file:

{
  "mcpServers": {
    "civasure": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote",
        "https://api.civasure.com/mcp",
        "--header",
        "Authorization: Bearer civ_live_YOUR_KEY"
      ]
    }
  }
}

3. Restart the app (fully quit and reopen), then ask away.

The developer docs walk through this step by step, and the full API reference documents every field — including the plain REST API, if you’d rather pull the same data into a dashboard or report instead of an assistant.

Who this is for

If your team already works with AI assistants, MCP turns Civasure into a data source you can simply talk to — useful for quick situational questions, drafting outreach plans, or pulling a number into a briefing without breaking stride. If you’d rather not touch a config file, the dashboard does everything the tools do, visually.

Either way, the data underneath is the same: anonymous, scoped to your area, and built for action.

Get started: Read the docs → · Create a key in Settings → API