Diagnoses and resolves common ComfyUI execution errors, performance bottlenecks, and configuration issues.
The ComfyUI Troubleshooting skill empowers Claude with specialized knowledge to identify, debug, and fix workflow failures within the ComfyUI ecosystem. It provides actionable solutions for critical issues including CUDA Out of Memory (OOM) errors, missing custom nodes, tensor mismatches, and unexpected visual artifacts like black images. By following a systematic diagnosis strategy using execution history and logs, this skill helps users maintain stable and efficient generative AI pipelines.
Key Features
01Missing node identification and custom node installation guidance
02VRAM optimization strategies for solving CUDA Out of Memory errors
03Debugging checklists for black images, NaN tensors, and CLIP overflows
04Resolution of tensor dtype and device (CPU vs GPU) mismatches
05Systematic error diagnosis using execution history and system logs
0611 GitHub stars
Use Cases
01Troubleshooting silent rendering failures where workflows succeed but produce empty or black images
02Fixing workflow import errors by identifying and installing missing custom node dependencies
03Recovering from VRAM bottlenecks by implementing FP8 quantization or tiled VAE decoding
