RevitSync

Revit posts a bounding-box snapshot to a local ASP.NET Core API. A React / Three.js page polls that API and draws the same elements. Clicks in the browser enqueue commands. The add-in polls the queue and applies them on Revit’s UI thread.

Personal project. Unofficial, not Autodesk work. Three processes on one machine.

Revit add-in          ASP.NET Core API           Browser
Export / DocumentChanged
  POST /api/geometry  →  in-memory snapshot  →  GET /api/geometry/latest
Click / drag / delete
  GET /api/commands   ←  in-memory queue     ←  POST /api/commands

What it does

Revit to web:

• World-axis bounding boxes from 13 categories in the active view.
• DocumentChanged debounce, then an ExternalEvent re-export.
• Selection ids ride on the next snapshot. The viewer outlines them.

Web to Revit:

• ADD_BOXES creates tagged DirectShape boxes.
• MOVE_ELEMENT / DELETE_ELEMENTS apply only to those tagged DirectShapes.
• SELECT_ELEMENTS sets selection and zooms. No transaction.

Move and delete from the web do not apply to ordinary Revit walls, floors, or families.

What I built

• C# / .NET Framework 4.8 Revit 2026 add-in.
• ASP.NET Core 9 API with in-memory snapshot and command queue.
• React / TypeScript / Three.js viewer with ETag / If-None-Match polling.

Limits

• Geometry is axis-aligned bounding boxes, not meshes.
• Snapshot and command queue live in process memory.
• No auth, no TLS, no multi-user, no conflict resolution.
• Polling, not WebSockets.
• Add-in targets .NET Framework 4.8 while Autodesk’s official Revit 2025/2026 add-in target is .NET 8.