Validate NUI Flow locally, then submit it through ui-runtime to the WGPU runtime for deployment.
MIT
524,660 tools. Updated 2026-09-06 17:35
"wgpu" matching MCP tools:
- Run Python code on Google Colab with GPU/TPU acceleration. Collects generated files (images, CSVs, models) by zipping and downloading them to a local directory.MIT
- List all background jobs on your GPU runtime to find job IDs for polling results or canceling tasks.MIT
- Execute a local .py script on a Google Colab GPU/TPU runtime. Supports T4, L4, A100, H100 accelerators.MIT
- Execute Python code on Google Colab GPU/TPU runtimes, with support for multiple accelerators, background jobs, and Google Drive file integration.MIT
- Check server status to see available accelerators and current job state, then choose an accelerator or wait for a running job before executing code.MIT
Matching MCP Servers
- AlicenseAqualityCmaintenanceEnables MCP-compatible AI assistants to run Python code on Google Colab GPU/TPU runtimes, supporting accelerators like T4, A100, H100, with background execution and Google Drive integration.102MIT
- AlicenseAqualityFmaintenanceUnified AI compute API gateway for agents. Access 30+ services and 95+ models across 8 backends (Groq, Together AI, DeepInfra, Fireworks, Replicate, RunPod) with native x402 USDC payments, Stripe, and crypto top-up.5592MIT
Matching MCP Connectors
Live GPU rental market: 2,500+ offers across a dozen provider feeds. History, watches, limit orders.
Agent-to-agent network for teams: dm, who-knows-X routing, shared rooms. Human-in-the-loop.
- Cancel an active background job to release Colab GPU resources. Use colab_jobs to find job IDs, then cancel and verify with colab_poll.MITDestructive
- Poll a background job for current status and results. Retrieve execution output when complete or error details on failure.MIT
- Downloads a file from Google Drive to a local path. Retrieve Colab execution results or any Drive file by specifying the Drive path and local destination.MIT
- Unload all models from VRAM and clear CUDA cache in ComfyUI to free GPU memory for new tasks.MIT
- Verify the MCP server version on Google Colab GPU to ensure compatibility or debug issues.MIT
- Verify ComfyUI is running and responsive, and check if the GPU optimizer is active.MIT
- Monitor GPU VRAM usage, utilization, temperature, and power draw to optimize performance.MIT
- Free up GPU memory by offloading a model from GPU to CPU. Specify which model to offload using its index, or offload the largest model with -1.MIT
- Frees GPU VRAM by terminating all compute processes. Use when ComfyUI runs out of memory.MIT
- Upload a local file to a specified Google Drive folder to stage data for Colab GPU runtimes. Use with colab_execute's drive_fetch parameter.MIT
- Get real-time GPU metrics including device info, VRAM usage, utilization, and temperature for all NVIDIA GPUs.MIT
- Browse available AI compute services with pricing and model details to select GPU resources for AI applications.MIT
- Automatically free GPU VRAM by offloading largest models to CPU until target free memory is reached. Specify target_free_mb or use auto mode.MIT
- Set the VRAM threshold between 50-95% to automatically offload models to CPU when memory exceeds the limit, preventing out-of-memory errors.MIT
- Estimate GPU compute service costs before execution to budget AI workloads across multiple providers.MIT