Enables AI agents to perform automated Blender UI testing and user-story validation through an isolated virtual machine, with tools for computer-use input, framebuffer screenshots, UI Automation trees, Blender bpy telemetry, scene state inspection, and snapshot rollback.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@blender-cu-vm-mcp Open Blender, add a cube via the UI menu, then screenshot the viewport"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
An isolated, GPU-partitioned local Windows 11 Virtual Machine environment connected via Model Context Protocol (MCP) specifically engineered for AI coding agents to perform automated Blender UI testing, user-story validation, and Computer-Use workflows without disturbing your host desktop.
+---------------------------------------------------------------------------------------------+
| Coding Agents (Claude Code / OMP / Codex) |
+---------------------------------------------------------------------------------------------+
│
│ (MCP JSON-RPC via stdio)
▼
+─────────────────────────────────────────────────────────────────────────────────────────────+
| Host Layer: `blender-cu-vm-mcp` Server |
| - MCP Tool Handler (Computer Use, Analytical Inspection, Blender Telemetry, File Staging) |
| - VM Lifecycle Manager (Hyper-V Socket / PowerShell WMI / Sub-2s Snapshot Reset) |
+─────────────────────────────────────────────────────────────────────────────────────────────+
│
┌───────────────────────┴───────────────────────┐
│ Hyper-V Socket (HV-SOCK) / Internal VMSwitch │
▼ ▼
+───────────────────────────────────────────────+ +─────────────────────────────────────────+
| Windows 11 Guest VM (Hyper-V) | | Secondary (Fast CI): WSL2 Container |
| - NVIDIA GPU-PV (RTX 4080 Super vGPU: 8GB) | | - Mesa / Direct3D 12 GPU Accel |
| - Virtual Display Driver (1080p Fixed 60Hz) | | - Virtual X11 Display (Xvfb/Weston) |
| - Guest Agent Daemon (FastAPI / gRPC) | | - Guest Agent Daemon (Linux) |
| - In-Process Blender Telemetry Bridge (bpy) | | - Blender bpy IPC Bridge |
+───────────────────────────────────────────────+ +─────────────────────────────────────────+Related MCP server: WindowsPC-MCP
Zero Host Disturbance: Synthetic mouse movements, drags, clicks, and keystrokes are executed exclusively inside the guest OS. Your real cursor and window focus remain completely untouched.
Hardware-Accelerated GPU Rendering: NVIDIA GPU-PV gives the guest VM near-native access to the host RTX 4080 Super (DirectX 12, Vulkan, OpenGL, and CUDA/OptiX).
Deterministic 1080p Display: Open-source Virtual Display Driver (IddSampleDriver) locks the virtual screen to 1920x1080 @ 60Hz with 100% (96 DPI) scaling—preventing coordinate drift and sleeping monitors.
Hybrid Multimodal Feedback: Agents receive:
Visual: Framebuffer screenshots with optional coordinate grid overlays.
Analytical: Windows UI Automation tree, bounding boxes, and window responsiveness.
Deep Telemetry: Blender bpy state, active modifiers, node tree connections, and real-time stdout/stderr logs.
Sub-2-Second State Rollback: Fast Hyper-V snapshot restoration resets the VM to a clean golden base after destructive or experimental runs.
blender-cu-vm/
├── host/ # Host MCP Server & Hyper-V Controller
│ ├── mcp_server.py # Stdio JSON-RPC MCP server with 14 tools
│ ├── hv_transport.py # Hyper-V Socket (AF_HYPERV) & HTTP transport
│ ├── vm_controller.py # PowerShell WMI lifecycle & snapshot manager
│ └── asset_sync.py # Bi-directional file and addon staging
├── guest/ # Guest Agent Daemon (runs inside VM)
│ ├── guest_daemon.py # FastAPI HTTP/HV-SOCK unified server
│ ├── screen_capture.py # DXGI Desktop Duplication & visual diffs
│ ├── input_controller.py # Win32 SendInput (clicks, drags, typing)
│ ├── ui_automation.py # Windows UI Automation tree inspector
│ └── video_recorder.py # Hardware-accelerated NVENC MP4 recorder
├── blender/ # Blender Embedded Runtime Bridge
│ ├── cu_telemetry_bridge.py # Non-blocking TCP telemetry server
│ ├── crash_interceptor.py # C-level stdout/stderr stream tee
│ └── state_inspector.py # Declarative scene invariant checker
├── scripts/ # Automated Setup & Provisioning
│ ├── setup_vm_gpupv.ps1 # Automated Hyper-V Gen2 VM creator
│ ├── stage_gpupv_drivers.ps1 # NVIDIA GPU-PV driver packaging & injection
│ ├── setup_virtual_display.ps1 # Virtual display & autologon configuration
│ └── manage_golden_snapshot.ps1 # Instant snapshot creation & rollback
├── tests/ # Verification & E2E Test Suite
│ ├── test_blender_user_story.py # 12-stage automated test suite
│ └── verify_isolation.py # Zero host disturbance verification
├── mcp-config.json # Registration snippet for Claude Code / OMP
└── README.mdProvisions the VM if missing, builds the unattended bootstrap ISO
(scripts\build_bootstrap_media.ps1, IMAPI2 — no ADK needed), attaches it
alongside the Windows 11 install ISO, boots the VM, and polls the guest
daemon health endpoint until the fully-unattended chain
(OOBE → auto-logon → bootstrap.cmd → install_guest.ps1 → daemon on
192.168.122.100:8000) completes:
cd C:\tmp\blender-cu-vm
.\scripts\deploy_guest_os.ps1 -WindowsISO "C:\ISOs\Win11_24H2_English_x64.iso"
# Watch graphically at any time: vmconnect.exe localhost Blender-CU-VMMirror the host RTX 4080 Super driver binaries into the guest. Use Stage
before install, or OnlineCopy once the guest OS is running:
.\scripts\stage_gpupv_drivers.ps1 -VMName "Blender-CU-VM" -Mode "OnlineCopy"The unattended chain above already runs guest\install_guest.ps1 (static IP
192.168.122.100, GPU-PV drivers if staged, 1080p virtual display lock,
hidden daemon startup). To run it by hand inside the guest instead:
# From the bootstrap DVD drive (e.g. E:) or C:\BlenderCU\guest:
powershell -ExecutionPolicy Bypass -File .\install_guest.ps1.\manage_golden_snapshot.ps1 -VMName "Blender-CU-VM" -SnapshotName "golden_base" -Action "Create"Add the following to your ~/.claude.json or ~/.omp/agent/config.yml:
{
"mcpServers": {
"blender-cu-vm": {
"command": "python",
"args": [
"C:\\tmp\\blender-cu-vm\\host\\mcp_server.py"
],
"env": {
"BLENDER_VM_NAME": "Blender-CU-VM",
"BLENDER_GUEST_URL": "http://192.168.122.100:8000"
}
}
}
}To verify all subsystems and run the simulated Blender user story:
python blender-cu-vm/tests/test_blender_user_story.py
python blender-cu-vm/tests/verify_isolation.pyThis server cannot be deployed
Cloud Blender for AI agents: scenes, assets, renders, MP4, STL, GLB — over hosted remote MCP.
Eyes and hands on real Windows PCs — observe, click, type via Glasswarp API.
Validates AI infra code on real VMs. Self-corrects until it works. No containers, no sandboxes.
Operate Linux, macOS and Windows from your LLM. Every action runs through an auditable allowlist.
