mckruz/comfyui-expert
15 skills8.2k installs1.3k starsGitHub
Install
npx skills add https://github.com/mckruz/comfyui-expertSkills in this repo
1Comfyui Apicomfyui-api is an agent skill from mckruz/comfyui-expert that connect to a running comfyui instance, queue workflows, monitor execution, and retrieve results. supports both online (rest api) and offline (json export) modes. use when executing comfyui workflows o. # ComfyUI API Skill Connect to ComfyUI's REST API to execute workflows, monitor progress, and retrieve outputs. ## Configuration - **Default URL**: `http://127.0.0.1:8188` - **Custom URL**: Set in project manifest or pass as parameter - **Timeout**: 30s for API calls, no timeout for generation polling ## Two Modes ### Online Mode (ComfyUI Runn Developers invoke comfyui-api during build/integrations work for generative media tasks. The skill documents triggers, prerequisites, and step-by-step workflows grounded in SKILL.md. Compatible with Claude Code, Cursor, and Codex agent runtimes that load marketplace skills.1.5kinstalls2Comfyui Workflow BuilderThe comfyui-workflow- skill translates natural language requests into executable ComfyUI workflow JSON with correct class_types, connections, and output indices. It reads state/inventory.json for available checkpoints, LoRAs, ControlNet models, and custom nodes before selecting pipeline patterns for txt2img, img2img, inpainting, identity preservation, video, upscale, and FaceDetailer flows. Workflow JSON uses string node IDs with array connection format source_id and output_index. Validation checks class_type existence, model filenames, required connections, VRAM estimates, and resolution compatibility such as 1024 for FLUX and SDXL. Use when developers need ComfyUI node graphs from prompts without installation or custom node development guidance.1.5kinstalls3Comfyui Video Pipelinecomfyui-video-pipeline is a user-invocable Claude Code skill from mckruz/comfyui-expert that routes video requests to one of three ComfyUI engines based on quality needs and hardware. The readme documents engine selection logic, including film-level quality paths that assume 24GB+ VRAM, and supports text-to-video, image-to-video, talking heads, and motion-controlled animation from character images or text descriptions. Metadata requires curl or wget and uses COMFYUI_URL as the primary environment variable across darwin, linux, and win32. Developers reach for comfyui-video-pipeline when they already run ComfyUI and want an agent to pick Wan 2.2, FramePack, or AnimateDiff instead of hand-building workflows per request. The skill is an orchestration layer over ComfyUI graphs, not a standalone video encoder or cloud API wrapper.880installs4Comfyui Prompt Engineercomfyui-prompt-engineer is a user-invocable skill from mckruz/comfyui-expert that crafts prompts optimized for specific ComfyUI checkpoints and identity methods. It documents distinct rules for FLUX.1 dev/schnell/Kontext (natural language, CFG 3.5-4), SDXL, SD1.5, and Wan video models, including quality tags, negative prompts, and syntax conventions each architecture expects. Developers reach for comfyui-prompt-engineer when ComfyUI workflows produce flat or off-model outputs because generic prompt templates ignore checkpoint-specific behavior. The skill is OpenClaw-compatible across darwin, linux, and win32 and focuses on translating creative intent into prompts that match the target model's response profile.874installs5Comfyui Troubleshootercomfyui-troubleshooter is a user-invocable agent skill in mckruz/comfyui-expert for diagnosing and fixing ComfyUI image-generation failures. It classifies issues into four categories—server errors such as connection refused and timeouts, workflow errors from missing custom nodes or broken graphs, quality problems in outputs, and performance issues including CUDA and VRAM exhaustion—then suggests fixes from error patterns, dependency gaps, and community-known workarounds. The skill supports darwin, linux, and win32 environments where ComfyUI workflows drive Stable Diffusion pipelines. Developers reach for comfyui-troubleshooter when ComfyUI crashes on launch, custom nodes fail to load, CUDA out-of-memory errors appear, or generated images look wrong despite a previously working graph.645installs6Comfyui Character Gencomfyui-character-gen is a ComfyUI expert skill from mckruz/comfyui-expert that helps developers and technical artists design node graphs for consistent character generation. It guides wiring ControlNet, IP-Adapter, LoRA, and checkpoint nodes so the same character identity survives pose changes, outfit swaps, and style variants across batches. Developers reach for comfyui-character-gen when building game sprite pipelines, comic panel workflows, or marketing asset factories that need repeatable faces, body proportions, and palette lock-in instead of one-off prompt roulette. The skill fits agent sessions where the deliverable is a ComfyUI workflow JSON—nodes, connections, and parameter defaults—rather than raw PNG output alone. It emphasizes graph structure for batch consistency, seed and reference-image strategy, and modular subgraphs reusable across pose and expression variants in local or studio ComfyUI installs.401installs7Comfyui InventoryComfyUI inventory from mckruz/comfyui-expert helps Claude catalog ComfyUI nodes, models, extensions, and workflows—keeping generative pipelines organized, reproducible, and ready for agent-driven image and media automation.346installs8Comfyui Lora TrainingGuides ComfyUI LoRA training end to end: dataset preparation, hyperparameter tuning, checkpoint validation, and export so teams can produce custom diffusion adapters for branded visuals, characters, or product imagery inside existing node workflows.339installs9Comfyui ResearchComfyUI research from mckruz/comfyui-expert equips Claude to investigate ComfyUI ecosystems—surveying nodes, workflows, models, and community techniques to inform early generative media and agent pipeline decisions before implementation.255installs10Comfyui Video ProductionCovers ComfyUI video production workflows including image-to-video models, motion guidance, temporal consistency, upscaling, clip assembly, and queued batch rendering for social campaigns, product demos, and automated agent content factories.255installs11Comfyui Prompt InterviewComfyUI Prompt Interview skill structures conversational prompt discovery for generative image and video workflows. It guides agents through subject, style, lighting, composition, and negative constraints, then maps responses into ComfyUI node parameters for reliable, repeatable creative output without guesswork.246installs12Comfyui Voice PipelineComfyUI Voice Pipeline skill covers assembling speech-to-text, text-to-speech, and audio timing nodes inside ComfyUI for narrated visuals and talking avatars. It documents graph wiring, model choices, sync strategies, and queue-friendly batch audio generation for multimodal creative automation.240installs13Video AssemblyExpert guidance for assembling finished videos inside ComfyUI by chaining generation, editing, and export nodes into stable workflows for AI-native content production.237installs14Project ManagerComfyUI Project Manager skill helps teams plan and track generative media productions: milestones, asset manifests, workflow versions, dependencies, and handoffs. It keeps multi-contributor ComfyUI graphs organized so complex image or video batches ship on schedule with traceable changes.227installs15Video PublisherGuides publishing of videos produced through ComfyUI expert workflows, covering export settings, destination configuration, and automated distribution so rendered generative clips reach platforms without manual upload steps.224installs