AEC Model Bridge
MCP serverCommunity
A Python server and Revit 2024–2027 add-in where every model change waits for a plan you approve in a Revit panel. By Sam-AEC.
118 starsUpdated 5 days agoGitHub data from 5 Nov 2026
At a glance
- Type
- MCP server
- Works with
- Revit, Rhino 3D & Grasshopper
- How it connects
- Drives a running copy of the tool
- Access
- Read-only or read-write (configurable)
- Tool versions
- Revit 2024 to 2027 (add-in built in CI; the README says no behaviour has yet been checked in a live Revit session)
- Operating systems
- Windows
- Needs a Revit licence
- Yes
- Transport
- Local (stdio)
- AI clients documented
- Claude Desktop, VS Code (GitHub Copilot), Cursor, Codex
- Licence
- GPL-3.0-or-later
- Language
- Python
What it can do
- Reads warnings, elements and parameters straight away, and stages edits as a plan that you approve in the Revit panel before it runs.
- Covers about 100 Revit tools for elements, views, sheets, schedules, exports and worksharing, plus QA/QC and report modules that work from saved snapshots.
- The same hub reads IFC files through IfcOpenShell without Revit, and drives Rhino and Grasshopper and Speckle through separate providers.
- A mock mode lists every tool and returns sample responses on any machine without Revit.
Limitations
- Pre-release. The README says the add-in has only been compiled for each Revit year, not run against a live model, so treat Revit behaviour as unconfirmed and try it on a copy.
- Dual licensed under GPL-3.0-or-later with a Revit linking exception, or a commercial licence.
- Navisworks support is listed as in progress.
Install
{
"mcpServers": {
"aec-model-bridge": {
"command": "uvx",
"args": [
"--from",
"git+https://github.com/Sam-AEC/aec-model-bridge#subdirectory=packages/mcp-server-revit",
"aec-model-bridge"
],
"env": {
"MCP_REVIT_MODE": "bridge",
"MCP_REVIT_WORKSPACE_DIR": "C:\\RevitProjects",
"MCP_REVIT_ALLOWED_DIRECTORIES": "C:\\RevitProjects"
}
}
}
}MCP servers and skills run code on your computer. Review the source before installing.
AI-generated models, drawings and results must be checked by a qualified engineer before use.