magicplan MCP
MCP Server
The magicplan MCP server lets AI assistants like Claude and ChatGPT read your magicplan workspace and reason over it in a conversation. It implements the Model Context Protocol, the open standard for connecting AI clients to external data. Where the REST API is for your own software, the MCP server is for AI assistants.
What Can You Do with It?
Ask an assistant about your workspace in plain language and let it pull the data it needs:
-
Projects and floor plans — a project's details, the field data on its plan, measured areas by floor and room, and the forms captured on site.
-
Restoration documentation — the drying history from moisture instruments and the photos attached to each part of the building, localized to their room.
-
Files and estimates — attached files, and estimates down to individual line items and cost totals.
It is especially suited to property restoration, where claim details, moisture readings, photo evidence, and an estimate all need to be read together.
Before You Begin
Make sure you have:
-
A magicplan account with access to the workspace you want to connect.
-
Your workspace's Customer ID and API key, found under API & Integrations in your workspace settings (the same credentials used for the REST API).
-
An MCP-capable client, such as Claude or ChatGPT.
How to Connect
The server connects at:
https://cloud.magicplan.com/mcpThe steps depend on your client, but the flow is the same everywhere:
-
Add the connector. In your assistant's settings, add a custom connector (in Claude, under Settings → Connectors) and paste the URL above.
-
Authorize the workspace. Your assistant opens a magicplan sign-in page asking for your Customer ID and API key. Enter both and confirm. Your API key is never shared with the assistant.
-
Start working. Ask the assistant about a project by name and it takes it from there.
A connection acts as one workspace. To connect another, authorize again with that workspace's credentials.
Good to Know
-
The server is read-only — it never changes anything in your workspace.
-
Responses are shaped for AI, not for parity with the REST API. Large payloads like a full floor plan are compacted so they stay usable in a conversation.
-
Your assistant shows the current list of tools it can call — that list is the definitive reference for what is available.
Next Steps
Connect your workspace and ask your assistant to summarize one of your projects. For programmatic integrations, see the REST API reference.
