comfy plugin

• /comfy:batch

  No description in the file.

• /comfy:compare

  No description in the file.

• /comfy:convert

  No description in the file.

• /comfy:debug

  No description in the file.

• /comfy:director

  No description in the file.

• /comfy:gallery

  No description in the file.

• /comfy:gen

  No description in the file.

• /comfy:install

  No description in the file.

• /comfy:node-skill

  No description in the file.

• /comfy:recipe

  No description in the file.

• /comfy:viz

  No description in the file.

• @agent-comfy:debugger

  No description in the file.

• @agent-comfy:explorer

  No description in the file.

• @agent-comfy:optimizer

  No description in the file.

• @agent-comfy:researcher

  No description in the file.

• ai-toolkit-trainer

  Train custom LoRAs for WAN 2.2/2.1 video models and Z-Image models using a Node.js web UI and Python backend.

• anima-base

  Generates anime and manga illustrations using a 2B-parameter DiT model that supports Danbooru tags and natural language prompting.

• anima-lora-trainer

  Train custom anime LoRA adapters for the Anima diffusion model using a local Gradio interface and low VRAM settings.

• civitai

  Search for, download, and install AI models like LoRAs and checkpoints directly into ComfyUI for local image generation.

• color-correction

  Diagnose and fix video and image color issues objectively using measurable data points like black/white points, contrast, and saturation.

• comfyui-core

  Provides core knowledge for managing ComfyUI workflows, node types, data pipelines, and MCP tool integration for image generation.

• comfyui-frontend-extensions

  Author ComfyUI v2 frontend extensions using a typed, import-based API for nodes, widgets, and shell UI components.

• comfyui-launch-flags

  Select optimal startup flags for ComfyUI to manage VRAM, attention backends, and caching to prevent OOM errors and improve performance.

• comfyui-node-registry

  Author and publish ComfyUI custom node packs to the Comfy Registry using pyproject.toml and comfy-cli.

• debug-render

  Debug imperfect renders by using run-to-node execution to inspect intermediate latents, masks, and preprocessor maps.

• director

  Runs a complete short film production pipeline from text story to final video using AI image and video generation models.

• ernie-image

  Build text-to-image workflows using Baidu's ERNIE-Image models for precise multilingual text rendering, posters, and complex multi-object scenes.

• flux-txt2img

  Build text-to-image workflows using Flux.1 Dev, Flux 2 Klein 9B, and Turbo LoRA patterns with specific node configurations.

• ideogram-ultra

  Build local text-to-image and image-to-image workflows using open-weights models with structured JSON prompts for precise layout and text control.

• installer-packs

  Install model families, custom nodes, and workflows using manifest-driven installer packs.

• krea2-identity-edit

  Swaps clothing onto a specific person locally while preserving their face, pose, and background.

• krea2-txt2img

  Build text-to-image workflows using Krea 2's 12B-parameter Diffusion Transformer with support for structured JSON prompting and turbo distillation.

• local-llm-free

  Run the ComfyUI agent locally and offline using fine-tuned Gemma 4 models via Ollama.

• lora-manager

  Allows an agent to author LoRA Manager nodes in ComfyUI using the LoRA Text Loader to bypass panel addition refusals.

• ltx-director

  Edit the video timeline of the LTX Director node by manipulating the hidden timeline_data JSON widget instead of using DOM buttons.

• ltxv2-video

  Build video generation workflows for Lightricks LTX-2 and LTX-2.3 models including text-to-video, image-to-video, and synchronized audio.

• minimax-h3-video

  Build local video workflows in ComfyUI using MiniMax H3 weights for text-to-video, image-to-video, and reference-to-video generation.

• model-compatibility

  Provides a compatibility matrix for Stable Diffusion, SDXL, Flux, SD3, and video models including settings for loaders, resolutions, and samplers.

• model-registry

  Provides curated download URLs and target directories for AI models including Flux, WAN, LTX, Qwen, and SDXL to support ComfyUI workflows.

• panel-node-pack-sync

  Synchronizes the ComfyUI sidebar panel node-pack with the orchestrator to prevent version drift and command failures.

• panel-operations

  Provides procedural instructions for managing subgraphs, merging workflows, and debugging complex node graphs within the ComfyUI sidebar panel.

• prompt-engineering

  ComfyUI prompt engineering knowledge covering CLIP text encoding syntax, weight modifiers, model-specific prompting strategies, and best practices

• qwen-image-edit

  Build image editing workflows using the Qwen2.5-VL vision-language model to modify images via natural language instructions.

• qwen-txt2img

  Build text-to-image workflows using Qwen Image 2512 models with integrated or separate component loading in ComfyUI.

• report-bug

  Manages user bug reports by explaining that issue reporting is currently disabled and providing workarounds instead of filing formal reports.

• rgthree

  Configure and author rgthree-comfy nodes including group bypassers, LoRA loaders, and context bundles within ComfyUI workflows.

• train-character-lora

  Train character-specific LoRAs locally on FLUX.1-dev using GPU Docker and the ostris ai-toolkit.

• triton-sageattention

  Install Triton and SageAttention to accelerate ComfyUI video workflows via quantized attention kernels and torch.compile.

• troubleshooting

  Provides troubleshooting strategies and fixes for common ComfyUI errors including OOM, missing nodes, dtype mismatches, and black images.

• video-extend

  Extends existing video clips temporally using Pusa 2.2 temporal flowmatching and WanVideoWrapper in ComfyUI.

• video-upscale

  Upscale and restore video using temporal-aware models like SeedVR2 or FlashVSR and perform frame interpolation with RIFE or FILM.

• wan-flf-video

  Build video generation workflows that create smooth transitions between a start and end image using WAN 2.2 I2V models.

• wan-multitalk

  Create audio-driven talking avatar videos by syncing a still portrait's lip movements and head motion to an audio track using WAN 2.1.

• wan-scail-replacement

  Replaces characters in driving videos with a reference image using the WAN 2.1 model and SAM3 masking.

• wan-t2v-video

  Build text-to-video workflows using a dual hi-lo model architecture with support for lightning LoRAs and VACE modules.

+2

• hooks

  No description in the file.

• mcp

  No description in the file.