[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7DACFKKCGS4AE87EMD99JWG":3,"skill-same:01M7DACFKKCGS4AE87EMD99JWG":210},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":54,"commitSha":55,"skillDirectory":56,"forksCount":14,"repositoryHtmlUrl":57,"repositoryHomepageUrl":6,"ownerHtmlUrl":58,"ownerBlogUrl":59,"ownerDescription":60,"updatedAt":61,"install":62,"resources":75},"01M7DACFKKCGS4AE87EMD99JWG","comfyui",null,"Generate images, video, and audio with ComfyUI — install, launch, manage nodes\u002Fmodels, run workflows with parameter injection. Uses the official comfy-cli for lifecycle and direct REST\u002FWebSocket API for execution.","GrupoANDevelopment-m\u002Falberto-ai","GrupoANDevelopment-m","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F257455179?v=4","01M7D69X2N2KEVN0SJF6WGM6SH","Antônio  Ndombe","01M7D69X3P5JTT8H8S904N3888",0,"MIT","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: comfyui\ndescription: \"Generate images, video, and audio with ComfyUI — install, launch, manage nodes\u002Fmodels, run workflows with parameter injection. Uses the official comfy-cli for lifecycle and direct REST\u002FWebSocket API for execution.\"\nversion: 5.1.0\nauthor: [kshitijk4poor, alt-glitch, purzbeats]\nlicense: MIT\nplatforms: [macos, linux, windows]\ncompatibility: \"Requires ComfyUI (local, Comfy Desktop, or Comfy Cloud) and comfy-cli (auto-installed via pipx\u002Fuvx by the setup script).\"\nprerequisites:\n  commands: [\"python3\"]\nsetup:\n  help: \"Run scripts\u002Fhardware_check.py FIRST to decide local vs Comfy Cloud; then scripts\u002Fcomfyui_setup.sh auto-installs locally (or use Cloud API key for platform.comfy.org).\"\nmetadata:\n  hermes:\n    tags:\n      - comfyui\n      - image-generation\n      - stable-diffusion\n      - flux\n      - sd3\n      - wan-video\n      - hunyuan-video\n      - creative\n      - generative-ai\n      - video-generation\n    related_skills: [stable-diffusion-image-generation, image_gen]\n    category: creative\n---\n\n# ComfyUI\n\nGenerate images, video, audio, and 3D content through ComfyUI using the\nofficial `comfy-cli` for setup\u002Flifecycle and direct REST\u002FWebSocket API\nfor workflow execution.\n\n## What's in this skill\n\n**Reference docs (`references\u002F`):**\n\n- `official-cli.md` — every `comfy ...` command, with flags\n- `rest-api.md` — REST + WebSocket endpoints (local + cloud), payload schemas\n- `workflow-format.md` — API-format JSON, common node types, param mapping\n- `template-integrity.md` — converting `comfyui-workflow-templates` from\n  editor format to API format: Reroute bypass, dotted dynamic-input keys\n  (`values.a`, `resize_type.width`), Cloud quirks (302 redirect, 1 concurrent\n  free-tier job, 1080p VRAM ceiling), Discord-compatible ffmpeg stitch.\n  Authored by [@purzbeats](https:\u002F\u002Fgithub.com\u002Fpurzbeats). Load this whenever\n  you're starting from an official template.\n\n**Scripts (`scripts\u002F`):**\n\n| Script | Purpose |\n|--------|---------|\n| `_common.py` | Shared HTTP, cloud routing, node catalogs (don't run directly) |\n| `hardware_check.py` | Probe GPU\u002FVRAM\u002Fdisk → recommend local vs Comfy Cloud |\n| `comfyui_setup.sh` | Hardware check + comfy-cli + ComfyUI install + launch + verify |\n| `extract_schema.py` | Read a workflow → list controllable params + model deps |\n| `check_deps.py` | Check workflow against running server → list missing nodes\u002Fmodels |\n| `auto_fix_deps.py` | Run check_deps then `comfy node install` \u002F `comfy model download` |\n| `run_workflow.py` | Inject params, submit, monitor, download outputs (HTTP or WS) |\n| `run_batch.py` | Submit a workflow N times with sweeps, parallel up to your tier |\n| `ws_monitor.py` | Real-time WebSocket viewer for executing jobs (live progress) |\n| `health_check.py` | Verification checklist runner — comfy-cli + server + models + smoke test |\n| `fetch_logs.py` | Pull traceback \u002F status messages for a given prompt_id |\n\n**Example workflows (`workflows\u002F`):** SD 1.5, SDXL, Flux Dev, SDXL img2img,\nSDXL inpaint, ESRGAN upscale, AnimateDiff video, Wan T2V. See\n`workflows\u002FREADME.md`.\n\n## When to Use\n\n- User asks to generate images with Stable Diffusion, SDXL, Flux, SD3, etc.\n- User wants to run a specific ComfyUI workflow file\n- User wants to chain generative steps (txt2img → upscale → face restore)\n- User needs ControlNet, inpainting, img2img, or other advanced pipelines\n- User asks to manage ComfyUI queue, check models, or install custom nodes\n- User wants video\u002Faudio\u002F3D generation via AnimateDiff, Hunyuan, Wan, AudioCraft, etc.\n\n## Architecture: Two Layers\n\n```\n┌─────────────────────────────────────────────────────┐\n│ Layer 1: comfy-cli (official lifecycle tool)        │\n│   Setup, server lifecycle, custom nodes, models     │\n│   → comfy install \u002F launch \u002F stop \u002F node \u002F model    │\n└─────────────────────────┬───────────────────────────┘\n                          │\n┌─────────────────────────▼───────────────────────────┐\n│ Layer 2: REST\u002FWebSocket API + skill scripts         │\n│   Workflow execution, param injection, monitoring   │\n│   POST \u002Fapi\u002Fprompt, GET \u002Fapi\u002Fview, WS \u002Fws           │\n│   → run_workflow.py, run_batch.py, ws_monitor.py    │\n└─────────────────────────────────────────────────────┘\n```\n\n**Why two layers?** The official CLI is excellent for installation and server\nmanagement but has minimal workflow execution support. The REST\u002FWS API fills\nthat gap — the scripts handle param injection, execution monitoring, and\noutput download that the CLI doesn't do.\n\n## Quick Start\n\n### Detect environment\n\n```bash\n# What's available?\ncommand -v comfy >\u002Fdev\u002Fnull 2>&1 && echo \"comfy-cli: installed\"\ncurl -s http:\u002F\u002F127.0.0.1:8188\u002Fsystem_stats 2>\u002Fdev\u002Fnull && echo \"server: running\"\n\n# Can this machine run ComfyUI locally? (GPU\u002FVRAM\u002Fdisk check)\npython3 scripts\u002Fhardware_check.py\n```\n\nIf nothing is installed, see **Setup & Onboarding** below — but always run the\nhardware check first.\n\n### One-line health check\n\n```bash\npython3 scripts\u002Fhealth_check.py\n# → JSON: comfy_cli on PATH? server reachable? at least one checkpoint? smoke-test passes?\n```\n\n## Core Workflow\n\n### Step 1: Get a workflow JSON in API format\n\nWorkflows must be in API format (each node has `class_type`). They come from:\n\n- ComfyUI web UI → **Workflow → Export (API)** (newer UI) or\n  the legacy \"Save (API Format)\" button (older UI)\n- This skill's `workflows\u002F` directory (ready-to-run examples)\n- Community downloads (civitai, Reddit, Discord) — usually editor format,\n  must be loaded into ComfyUI then re-exported\n\nEditor format (top-level `nodes` and `links` arrays) is **not directly\nexecutable**. The scripts detect this and tell you to re-export.\n\n### Step 2: See what's controllable\n\n```bash\npython3 scripts\u002Fextract_schema.py workflow_api.json --summary-only\n# → {\"parameter_count\": 12, \"has_negative_prompt\": true, \"has_seed\": true, ...}\n\npython3 scripts\u002Fextract_schema.py workflow_api.json\n# → full schema with parameters, model deps, embedding refs\n```\n\n### Step 3: Run with parameters\n\n```bash\n# Local (defaults to http:\u002F\u002F127.0.0.1:8188)\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflow_api.json \\\n  --args '{\"prompt\": \"a beautiful sunset over mountains\", \"seed\": -1, \"steps\": 30}' \\\n  --output-dir .\u002Foutputs\n\n# Cloud (export API key once; uses correct \u002Fapi routing automatically)\nexport COMFY_CLOUD_API_KEY=\"comfyui-...\"\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflow_api.json \\\n  --args '{\"prompt\": \"...\"}' \\\n  --host https:\u002F\u002Fcloud.comfy.org \\\n  --output-dir .\u002Foutputs\n\n# Real-time progress via WebSocket (requires `pip install websocket-client`)\npython3 scripts\u002Frun_workflow.py \\\n  --workflow flux_dev.json \\\n  --args '{\"prompt\": \"...\"}' \\\n  --ws\n\n# img2img \u002F inpaint: pass --input-image to upload + reference automatically\npython3 scripts\u002Frun_workflow.py \\\n  --workflow sdxl_img2img.json \\\n  --input-image image=.\u002Fphoto.png \\\n  --args '{\"prompt\": \"make it watercolor\", \"denoise\": 0.6}'\n\n# Batch \u002F sweep: 8 random seeds, parallel up to cloud tier limit\npython3 scripts\u002Frun_batch.py \\\n  --workflow sdxl.json \\\n  --args '{\"prompt\": \"abstract\"}' \\\n  --count 8 --randomize-seed --parallel 3 \\\n  --output-dir .\u002Foutputs\u002Fbatch\n```\n\n`-1` for `seed` (or omitting it with `--randomize-seed`) generates a fresh\nrandom seed per run.\n\n### Step 4: Present results\n\nThe scripts emit JSON to stdout describing every output file:\n\n```json\n{\n  \"status\": \"success\",\n  \"prompt_id\": \"abc-123\",\n  \"outputs\": [\n    {\"file\": \".\u002Foutputs\u002Fsdxl_00001_.png\", \"node_id\": \"9\",\n     \"type\": \"image\", \"filename\": \"sdxl_00001_.png\"}\n  ]\n}\n```\n\n## Decision Tree\n\n| User says | Tool | Command |\n|-----------|------|---------|\n| **Lifecycle (use comfy-cli)** | | |\n| \"install ComfyUI\" | comfy-cli | `bash scripts\u002Fcomfyui_setup.sh` |\n| \"start ComfyUI\" | comfy-cli | `comfy launch --background` |\n| \"stop ComfyUI\" | comfy-cli | `comfy stop` |\n| \"install X node\" | comfy-cli | `comfy node install \u003Cname>` |\n| \"download X model\" | comfy-cli | `comfy model download --url \u003Curl> --relative-path models\u002Fcheckpoints` |\n| \"list installed models\" | comfy-cli | `comfy model list` |\n| \"list installed nodes\" | comfy-cli | `comfy node show installed` |\n| **Execution (use scripts)** | | |\n| \"is everything ready?\" | script | `health_check.py` (optionally with `--workflow X --smoke-test`) |\n| \"what can I change in this workflow?\" | script | `extract_schema.py W.json` |\n| \"check if W's deps are met\" | script | `check_deps.py W.json` |\n| \"fix missing deps\" | script | `auto_fix_deps.py W.json` |\n| \"generate an image\" | script | `run_workflow.py --workflow W --args '{...}'` |\n| \"use this image\" (img2img) | script | `run_workflow.py --input-image image=.\u002Fx.png ...` |\n| \"8 variations with random seeds\" | script | `run_batch.py --count 8 --randomize-seed ...` |\n| \"show me live progress\" | script | `ws_monitor.py --prompt-id \u003Cid>` |\n| \"fetch the error from job X\" | script | `fetch_logs.py \u003Cprompt_id>` |\n| **Direct REST** | | |\n| \"what's in the queue?\" | REST | `curl http:\u002F\u002FHOST:8188\u002Fqueue` (local) or `--host https:\u002F\u002Fcloud.comfy.org` |\n| \"cancel that\" | REST | `curl -X POST http:\u002F\u002FHOST:8188\u002Finterrupt` |\n| \"free GPU memory\" | REST | `curl -X POST http:\u002F\u002FHOST:8188\u002Ffree` |\n\n## Setup & Onboarding\n\nWhen a user asks to set up ComfyUI, **the FIRST thing to do is ask whether\nthey want Comfy Cloud (hosted, zero install, API key) or Local (install\nComfyUI on their machine)**. Don't start running install commands or hardware\nchecks until they've answered.\n\n**Official docs:** https:\u002F\u002Fdocs.comfy.org\u002Finstallation\n**CLI docs:** https:\u002F\u002Fdocs.comfy.org\u002Fcomfy-cli\u002Fgetting-started\n**Cloud docs:** https:\u002F\u002Fdocs.comfy.org\u002Fget_started\u002Fcloud\n**Cloud API:** https:\u002F\u002Fdocs.comfy.org\u002Fdevelopment\u002Fcloud\u002Foverview\n\n### Step 0: Ask Local vs Cloud (ALWAYS FIRST)\n\nSuggested script:\n\n> \"Do you want to run ComfyUI locally on your machine, or use Comfy Cloud?\n>\n> - **Comfy Cloud** — hosted on RTX 6000 Pro GPUs, all common models pre-installed,\n>   zero setup. Requires an API key (paid subscription required to actually run\n>   workflows; free tier is read-only). Best if you don't have a capable GPU.\n> - **Local** — free, but your machine MUST meet the hardware requirements:\n>   - NVIDIA GPU with **≥6 GB VRAM** (≥8 GB for SDXL, ≥12 GB for Flux\u002Fvideo), OR\n>   - AMD GPU with ROCm support (Linux), OR\n>   - Apple Silicon Mac (M1+) with **≥16 GB unified memory** (≥32 GB recommended).\n>   - Intel Macs and machines with no GPU will NOT work — use Cloud instead.\n>\n> Which would you like?\"\n\nRouting:\n\n- **Cloud** → skip to **Path A**.\n- **Local** → run hardware check first, then pick a path from Paths B–E based on the verdict.\n- **Unsure** → run the hardware check and let the verdict decide.\n\n### Step 1: Verify Hardware (ONLY if user chose local)\n\n```bash\npython3 scripts\u002Fhardware_check.py --json\n# Optional: also probe `torch` for actual CUDA\u002FMPS:\npython3 scripts\u002Fhardware_check.py --json --check-pytorch\n```\n\n| Verdict    | Meaning                                                       | Action |\n|------------|---------------------------------------------------------------|--------|\n| `ok`       | ≥8 GB VRAM (discrete) OR ≥32 GB unified (Apple Silicon)       | Local install — use `comfy_cli_flag` from report |\n| `marginal` | SD1.5 works; SDXL tight; Flux\u002Fvideo unlikely                  | Local OK for light workflows, else **Path A (Cloud)** |\n| `cloud`    | No usable GPU, \u003C6 GB VRAM, \u003C16 GB Apple unified, Intel Mac, Rosetta Python | **Switch to Cloud** unless user explicitly forces local |\n\nThe script also surfaces `wsl: true` (WSL2 with NVIDIA passthrough) and\n`rosetta: true` (x86_64 Python on Apple Silicon — must reinstall as ARM64).\n\nIf verdict is `cloud` but the user wants local, do not proceed silently.\nShow the `notes` array verbatim and ask whether they want to (a) switch to\nCloud or (b) force a local install (will OOM or be unusably slow on modern models).\n\n### Choosing an Installation Path\n\nUse the hardware check first. The table below is the fallback for when the\nuser has already told you their hardware:\n\n| Situation | Recommended Path |\n|-----------|------------------|\n| `verdict: cloud` from hardware check | **Path A: Comfy Cloud** |\n| No GPU \u002F want to try without commitment | **Path A: Comfy Cloud** |\n| Windows + NVIDIA + non-technical | **Path B: ComfyUI Desktop** |\n| Windows + NVIDIA + technical | **Path C: Portable** or **Path D: comfy-cli** |\n| Linux + any GPU | **Path D: comfy-cli** (easiest) |\n| macOS + Apple Silicon | **Path B: Desktop** or **Path D: comfy-cli** |\n| Headless \u002F server \u002F CI \u002F agents | **Path D: comfy-cli** |\n\nFor the fully automated path (hardware check → install → launch → verify):\n\n```bash\nbash scripts\u002Fcomfyui_setup.sh\n# Or with overrides:\nbash scripts\u002Fcomfyui_setup.sh --m-series --port=8190 --workspace=\u002Fdata\u002Fcomfy\n```\n\nIt runs `hardware_check.py` internally, refuses to install locally when the\nverdict is `cloud` (unless `--force-cloud-override`), picks the right\n`comfy-cli` flag, and prefers `pipx`\u002F`uvx` over global `pip` to avoid polluting\nsystem Python.\n\n---\n\n### Path A: Comfy Cloud (No Local Install)\n\nFor users without a capable GPU or who want zero setup. Hosted on RTX 6000 Pro.\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Fget_started\u002Fcloud\n\n1. Sign up at https:\u002F\u002Fcomfy.org\u002Fcloud\n2. Generate an API key at https:\u002F\u002Fplatform.comfy.org\u002Flogin\n3. Set the key:\n   ```bash\n   export COMFY_CLOUD_API_KEY=\"comfyui-xxxxxxxxxxxx\"\n   ```\n4. Run workflows:\n   ```bash\n   python3 scripts\u002Frun_workflow.py \\\n     --workflow workflows\u002Fflux_dev_txt2img.json \\\n     --args '{\"prompt\": \"...\"}' \\\n     --host https:\u002F\u002Fcloud.comfy.org \\\n     --output-dir .\u002Foutputs\n   ```\n\n**Pricing:** https:\u002F\u002Fwww.comfy.org\u002Fcloud\u002Fpricing\n**Concurrent jobs:** Free\u002FStandard 1, Creator 3, Pro 5. Free tier\n**cannot run workflows via API** — only browse models. Paid subscription\nrequired for `\u002Fapi\u002Fprompt`, `\u002Fapi\u002Fupload\u002F*`, `\u002Fapi\u002Fview`, etc.\n\n---\n\n### Path B: ComfyUI Desktop (Windows \u002F macOS)\n\nOne-click installer for non-technical users. Currently Beta.\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Finstallation\u002Fdesktop\n- **Windows (NVIDIA):** https:\u002F\u002Fdownload.comfy.org\u002Fwindows\u002Fnsis\u002Fx64\n- **macOS (Apple Silicon):** https:\u002F\u002Fcomfy.org\n\nLinux is **not supported** for Desktop — use Path D.\n\n---\n\n### Path C: ComfyUI Portable (Windows Only)\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Finstallation\u002Fcomfyui_portable_windows\n\nDownload from https:\u002F\u002Fgithub.com\u002Fcomfyanonymous\u002FComfyUI\u002Freleases, extract,\nrun `run_nvidia_gpu.bat`. Update via `update\u002Fupdate_comfyui_stable.bat`.\n\n---\n\n### Path D: comfy-cli (All Platforms — Recommended for Agents)\n\nThe official CLI is the best path for headless\u002Fautomated setups.\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Fcomfy-cli\u002Fgetting-started\n\n#### Install comfy-cli\n\n```bash\n# Recommended:\npipx install comfy-cli\n# Or use uvx without installing:\nuvx --from comfy-cli comfy --help\n# Or (if pipx\u002Fuvx unavailable):\npip install --user comfy-cli\n```\n\nDisable analytics non-interactively:\n```bash\ncomfy --skip-prompt tracking disable\n```\n\n#### Install ComfyUI\n\n```bash\ncomfy --skip-prompt install --nvidia              # NVIDIA (CUDA)\ncomfy --skip-prompt install --amd                 # AMD (ROCm, Linux)\ncomfy --skip-prompt install --m-series            # Apple Silicon (MPS)\ncomfy --skip-prompt install --cpu                 # CPU only (slow)\ncomfy --skip-prompt install --nvidia --fast-deps  # uv-based dep resolution\n```\n\nDefault location: `~\u002Fcomfy\u002FComfyUI` (Linux), `~\u002FDocuments\u002Fcomfy\u002FComfyUI`\n(macOS\u002FWin). Override with `comfy --workspace \u002Fcustom\u002Fpath install`.\n\n#### Launch \u002F verify\n\n```bash\ncomfy launch --background                       # background daemon on :8188\ncomfy launch -- --listen 0.0.0.0 --port 8190    # LAN-accessible custom port\ncurl -s http:\u002F\u002F127.0.0.1:8188\u002Fsystem_stats      # health check\n```\n\n---\n\n### Path E: Manual Install (Advanced \u002F Unsupported Hardware)\n\nFor Ascend NPU, Cambricon MLU, Intel Arc, or other unsupported hardware.\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Finstallation\u002Fmanual_install\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fcomfyanonymous\u002FComfyUI.git\ncd ComfyUI\npip install torch torchvision torchaudio --extra-index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu130\npip install -r requirements.txt\npython main.py\n```\n\n---\n\n### Post-Install: Download Models\n\n```bash\n# SDXL (general purpose, ~6.5 GB)\ncomfy model download \\\n  --url \"https:\u002F\u002Fhuggingface.co\u002Fstabilityai\u002Fstable-diffusion-xl-base-1.0\u002Fresolve\u002Fmain\u002Fsd_xl_base_1.0.safetensors\" \\\n  --relative-path models\u002Fcheckpoints\n\n# SD 1.5 (lighter, ~4 GB, good for 6 GB cards)\ncomfy model download \\\n  --url \"https:\u002F\u002Fhuggingface.co\u002Fstable-diffusion-v1-5\u002Fstable-diffusion-v1-5\u002Fresolve\u002Fmain\u002Fv1-5-pruned-emaonly.safetensors\" \\\n  --relative-path models\u002Fcheckpoints\n\n# Flux Dev fp8 (smaller variant, ~12 GB)\ncomfy model download \\\n  --url \"https:\u002F\u002Fhuggingface.co\u002FComfy-Org\u002Fflux1-dev\u002Fresolve\u002Fmain\u002Fflux1-dev-fp8.safetensors\" \\\n  --relative-path models\u002Fcheckpoints\n\n# CivitAI (set token first):\ncomfy model download \\\n  --url \"https:\u002F\u002Fcivitai.com\u002Fapi\u002Fdownload\u002Fmodels\u002F128713\" \\\n  --relative-path models\u002Fcheckpoints \\\n  --set-civitai-api-token \"YOUR_TOKEN\"\n```\n\nList installed: `comfy model list`.\n\n### Post-Install: Install Custom Nodes\n\n```bash\ncomfy node install comfyui-impact-pack             # popular utility pack\ncomfy node install comfyui-animatediff-evolved     # video generation\ncomfy node install comfyui-controlnet-aux          # ControlNet preprocessors\ncomfy node install comfyui-essentials              # common helpers\ncomfy node update all\ncomfy node install-deps --workflow=workflow.json   # install everything a workflow needs\n```\n\n### Post-Install: Verify\n\n```bash\npython3 scripts\u002Fhealth_check.py\n# → comfy_cli on PATH? server reachable? checkpoints? smoke test?\n\npython3 scripts\u002Fcheck_deps.py my_workflow.json\n# → are this workflow's nodes\u002Fmodels\u002Fembeddings installed?\n\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflows\u002Fsd15_txt2img.json \\\n  --args '{\"prompt\": \"test\", \"steps\": 4}' \\\n  --output-dir .\u002Ftest-outputs\n```\n\n## Image Upload (img2img \u002F Inpainting)\n\nThe simplest way is to use `--input-image` with `run_workflow.py`:\n\n```bash\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflows\u002Fsdxl_img2img.json \\\n  --input-image image=.\u002Fphoto.png \\\n  --args '{\"prompt\": \"make it cyberpunk\", \"denoise\": 0.6}'\n```\n\nThe flag uploads `photo.png`, then injects its server-side filename into\nwhatever schema parameter is named `image`. For inpainting, pass both:\n\n```bash\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflows\u002Fsdxl_inpaint.json \\\n  --input-image image=.\u002Fphoto.png \\\n  --input-image mask_image=.\u002Fmask.png \\\n  --args '{\"prompt\": \"fill with flowers\"}'\n```\n\nManual upload via REST:\n```bash\ncurl -X POST \"http:\u002F\u002F127.0.0.1:8188\u002Fupload\u002Fimage\" \\\n  -F \"image=@photo.png\" -F \"type=input\" -F \"overwrite=true\"\n# Returns: {\"name\": \"photo.png\", \"subfolder\": \"\", \"type\": \"input\"}\n\n# Cloud equivalent:\ncurl -X POST \"https:\u002F\u002Fcloud.comfy.org\u002Fapi\u002Fupload\u002Fimage\" \\\n  -H \"X-API-Key: $COMFY_CLOUD_API_KEY\" \\\n  -F \"image=@photo.png\" -F \"type=input\" -F \"overwrite=true\"\n```\n\n## Cloud Specifics\n\n- **Base URL:** `https:\u002F\u002Fcloud.comfy.org`\n- **Auth:** `X-API-Key` header (or `?token=KEY` for WebSocket)\n- **API key:** set `$COMFY_CLOUD_API_KEY` once and the scripts pick it up automatically\n- **Output download:** `\u002Fapi\u002Fview` returns a 302 to a signed URL; the scripts\n  follow it and strip `X-API-Key` before fetching from the storage backend\n  (don't leak the API key to S3\u002FCloudFront).\n- **Endpoint differences from local ComfyUI:**\n  - `\u002Fapi\u002Fobject_info`, `\u002Fapi\u002Fqueue`, `\u002Fapi\u002Fuserdata` — **403 on free tier**;\n    paid only.\n  - `\u002Fhistory` is renamed to `\u002Fhistory_v2` on cloud (the scripts route\n    automatically).\n  - `\u002Fmodels\u002F\u003Cfolder>` is renamed to `\u002Fexperiment\u002Fmodels\u002F\u003Cfolder>` on cloud\n    (the scripts route automatically).\n  - `clientId` in WebSocket is currently ignored — all connections for a\n    user receive the same broadcast. Filter by `prompt_id` client-side.\n  - `subfolder` is accepted on uploads but ignored — cloud has a flat namespace.\n- **Concurrent jobs:** Free\u002FStandard: 1, Creator: 3, Pro: 5. Extras queue\n  automatically. Use `run_batch.py --parallel N` to saturate your tier.\n\n## Queue & System Management\n\n```bash\n# Local\ncurl -s http:\u002F\u002F127.0.0.1:8188\u002Fqueue | python3 -m json.tool\ncurl -X POST http:\u002F\u002F127.0.0.1:8188\u002Fqueue -d '{\"clear\": true}'    # cancel pending\ncurl -X POST http:\u002F\u002F127.0.0.1:8188\u002Finterrupt                      # cancel running\ncurl -X POST http:\u002F\u002F127.0.0.1:8188\u002Ffree \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"unload_models\": true, \"free_memory\": true}'\n\n# Cloud — same paths under \u002Fapi\u002F, plus:\npython3 scripts\u002Ffetch_logs.py --tail-queue --host https:\u002F\u002Fcloud.comfy.org\n```\n\n## Pitfalls\n\n1. **API format required** — every script and the `\u002Fapi\u002Fprompt` endpoint expect\n   API-format workflow JSON. The scripts detect editor format (top-level\n   `nodes` and `links` arrays) and tell you to re-export via\n   \"Workflow → Export (API)\" (newer UI) or \"Save (API Format)\" (older UI).\n\n2. **Server must be running** — all execution requires a live server.\n   `comfy launch --background` starts one. Verify with\n   `curl http:\u002F\u002F127.0.0.1:8188\u002Fsystem_stats`.\n\n3. **Model names are exact** — case-sensitive, includes file extension.\n   `check_deps.py` does fuzzy matching (with\u002Fwithout extension and folder\n   prefix), but the workflow itself must use the canonical name. Use\n   `comfy model list` to discover what's installed.\n\n4. **Missing custom nodes** — \"class_type not found\" means a required node\n   isn't installed. `check_deps.py` reports which package to install;\n   `auto_fix_deps.py` runs the install for you.\n\n5. **Working directory** — `comfy-cli` auto-detects the ComfyUI workspace.\n   If commands fail with \"no workspace found\", use\n   `comfy --workspace \u002Fpath\u002Fto\u002FComfyUI \u003Ccommand>` or\n   `comfy set-default \u002Fpath\u002Fto\u002FComfyUI`.\n\n6. **Cloud free-tier API limits** — `\u002Fapi\u002Fprompt`, `\u002Fapi\u002Fview`, `\u002Fapi\u002Fupload\u002F*`,\n   `\u002Fapi\u002Fobject_info` all return 403 on free accounts. `health_check.py` and\n   `check_deps.py` handle this gracefully and surface a clear message.\n\n7. **Timeout for video\u002Faudio workflows** — auto-detected when an output node\n   is `VHS_VideoCombine`, `SaveVideo`, etc.; the default jumps from 300 s to\n   900 s. Override explicitly with `--timeout 1800`.\n\n8. **Path traversal in output filenames** — server-supplied filenames are\n   passed through `safe_path_join` to refuse anything escaping `--output-dir`.\n   Keep this protection on — workflows with custom save nodes can produce\n   arbitrary paths.\n\n9. **Workflow JSON is arbitrary code** — custom nodes run Python, so\n   submitting an unknown workflow has the same trust profile as `eval`.\n   Inspect workflows from untrusted sources before running.\n\n10. **Auto-randomized seed** — pass `seed: -1` in `--args` (or use\n    `--randomize-seed` and omit the seed) to get a fresh seed per run.\n    The actual seed is logged to stderr.\n\n11. **`tracking` prompt** — first run of `comfy` may prompt for analytics.\n    Use `comfy --skip-prompt tracking disable` to skip non-interactively.\n    `comfyui_setup.sh` does this for you.\n\n## Verification Checklist\n\nUse `python3 scripts\u002Fhealth_check.py` to run the whole list at once. Manual:\n\n- [ ] `hardware_check.py` verdict is `ok` OR the user explicitly chose Comfy Cloud\n- [ ] `comfy --version` works (or `uvx --from comfy-cli comfy --help`)\n- [ ] `curl http:\u002F\u002FHOST:PORT\u002Fsystem_stats` returns JSON\n- [ ] `comfy model list` shows at least one checkpoint (local) OR\n      `\u002Fapi\u002Fexperiment\u002Fmodels\u002Fcheckpoints` returns models (cloud)\n- [ ] Workflow JSON is in API format\n- [ ] `check_deps.py` reports `is_ready: true` (or only `node_check_skipped`\n      on cloud free tier)\n- [ ] Test run with a small workflow completes; outputs land in `--output-dir`\n","# ComfyUI\n\nGenerate images, video, audio, and 3D content through ComfyUI using the\nofficial `comfy-cli` for setup\u002Flifecycle and direct REST\u002FWebSocket API\nfor workflow execution.\n\n## What's in this skill\n\n**Reference docs (`references\u002F`):**\n\n- `official-cli.md` — every `comfy ...` command, with flags\n- `rest-api.md` — REST + WebSocket endpoints (local + cloud), payload schemas\n- `workflow-format.md` — API-format JSON, common node types, param mapping\n- `template-integrity.md` — converting `comfyui-workflow-templates` from\n  editor format to API format: Reroute bypass, dotted dynamic-input keys\n  (`values.a`, `resize_type.width`), Cloud quirks (302 redirect, 1 concurrent\n  free-tier job, 1080p VRAM ceiling), Discord-compatible ffmpeg stitch.\n  Authored by [@purzbeats](https:\u002F\u002Fgithub.com\u002Fpurzbeats). Load this whenever\n  you're starting from an official template.\n\n**Scripts (`scripts\u002F`):**\n\n| Script | Purpose |\n|--------|---------|\n| `_common.py` | Shared HTTP, cloud routing, node catalogs (don't run directly) |\n| `hardware_check.py` | Probe GPU\u002FVRAM\u002Fdisk → recommend local vs Comfy Cloud |\n| `comfyui_setup.sh` | Hardware check + comfy-cli + ComfyUI install + launch + verify |\n| `extract_schema.py` | Read a workflow → list controllable params + model deps |\n| `check_deps.py` | Check workflow against running server → list missing nodes\u002Fmodels |\n| `auto_fix_deps.py` | Run check_deps then `comfy node install` \u002F `comfy model download` |\n| `run_workflow.py` | Inject params, submit, monitor, download outputs (HTTP or WS) |\n| `run_batch.py` | Submit a workflow N times with sweeps, parallel up to your tier |\n| `ws_monitor.py` | Real-time WebSocket viewer for executing jobs (live progress) |\n| `health_check.py` | Verification checklist runner — comfy-cli + server + models + smoke test |\n| `fetch_logs.py` | Pull traceback \u002F status messages for a given prompt_id |\n\n**Example workflows (`workflows\u002F`):** SD 1.5, SDXL, Flux Dev, SDXL img2img,\nSDXL inpaint, ESRGAN upscale, AnimateDiff video, Wan T2V. See\n`workflows\u002FREADME.md`.\n\n## When to Use\n\n- User asks to generate images with Stable Diffusion, SDXL, Flux, SD3, etc.\n- User wants to run a specific ComfyUI workflow file\n- User wants to chain generative steps (txt2img → upscale → face restore)\n- User needs ControlNet, inpainting, img2img, or other advanced pipelines\n- User asks to manage ComfyUI queue, check models, or install custom nodes\n- User wants video\u002Faudio\u002F3D generation via AnimateDiff, Hunyuan, Wan, AudioCraft, etc.\n\n## Architecture: Two Layers\n\n```\n┌─────────────────────────────────────────────────────┐\n│ Layer 1: comfy-cli (official lifecycle tool)        │\n│   Setup, server lifecycle, custom nodes, models     │\n│   → comfy install \u002F launch \u002F stop \u002F node \u002F model    │\n└─────────────────────────┬───────────────────────────┘\n                          │\n┌─────────────────────────▼───────────────────────────┐\n│ Layer 2: REST\u002FWebSocket API + skill scripts         │\n│   Workflow execution, param injection, monitoring   │\n│   POST \u002Fapi\u002Fprompt, GET \u002Fapi\u002Fview, WS \u002Fws           │\n│   → run_workflow.py, run_batch.py, ws_monitor.py    │\n└─────────────────────────────────────────────────────┘\n```\n\n**Why two layers?** The official CLI is excellent for installation and server\nmanagement but has minimal workflow execution support. The REST\u002FWS API fills\nthat gap — the scripts handle param injection, execution monitoring, and\noutput download that the CLI doesn't do.\n\n## Quick Start\n\n### Detect environment\n\n```bash\n# What's available?\ncommand -v comfy >\u002Fdev\u002Fnull 2>&1 && echo \"comfy-cli: installed\"\ncurl -s http:\u002F\u002F127.0.0.1:8188\u002Fsystem_stats 2>\u002Fdev\u002Fnull && echo \"server: running\"\n\n# Can this machine run ComfyUI locally? (GPU\u002FVRAM\u002Fdisk check)\npython3 scripts\u002Fhardware_check.py\n```\n\nIf nothing is installed, see **Setup & Onboarding** below — but always run the\nhardware check first.\n\n### One-line health check\n\n```bash\npython3 scripts\u002Fhealth_check.py\n# → JSON: comfy_cli on PATH? server reachable? at least one checkpoint? smoke-test passes?\n```\n\n## Core Workflow\n\n### Step 1: Get a workflow JSON in API format\n\nWorkflows must be in API format (each node has `class_type`). They come from:\n\n- ComfyUI web UI → **Workflow → Export (API)** (newer UI) or\n  the legacy \"Save (API Format)\" button (older UI)\n- This skill's `workflows\u002F` directory (ready-to-run examples)\n- Community downloads (civitai, Reddit, Discord) — usually editor format,\n  must be loaded into ComfyUI then re-exported\n\nEditor format (top-level `nodes` and `links` arrays) is **not directly\nexecutable**. The scripts detect this and tell you to re-export.\n\n### Step 2: See what's controllable\n\n```bash\npython3 scripts\u002Fextract_schema.py workflow_api.json --summary-only\n# → {\"parameter_count\": 12, \"has_negative_prompt\": true, \"has_seed\": true, ...}\n\npython3 scripts\u002Fextract_schema.py workflow_api.json\n# → full schema with parameters, model deps, embedding refs\n```\n\n### Step 3: Run with parameters\n\n```bash\n# Local (defaults to http:\u002F\u002F127.0.0.1:8188)\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflow_api.json \\\n  --args '{\"prompt\": \"a beautiful sunset over mountains\", \"seed\": -1, \"steps\": 30}' \\\n  --output-dir .\u002Foutputs\n\n# Cloud (export API key once; uses correct \u002Fapi routing automatically)\nexport COMFY_CLOUD_API_KEY=\"comfyui-...\"\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflow_api.json \\\n  --args '{\"prompt\": \"...\"}' \\\n  --host https:\u002F\u002Fcloud.comfy.org \\\n  --output-dir .\u002Foutputs\n\n# Real-time progress via WebSocket (requires `pip install websocket-client`)\npython3 scripts\u002Frun_workflow.py \\\n  --workflow flux_dev.json \\\n  --args '{\"prompt\": \"...\"}' \\\n  --ws\n\n# img2img \u002F inpaint: pass --input-image to upload + reference automatically\npython3 scripts\u002Frun_workflow.py \\\n  --workflow sdxl_img2img.json \\\n  --input-image image=.\u002Fphoto.png \\\n  --args '{\"prompt\": \"make it watercolor\", \"denoise\": 0.6}'\n\n# Batch \u002F sweep: 8 random seeds, parallel up to cloud tier limit\npython3 scripts\u002Frun_batch.py \\\n  --workflow sdxl.json \\\n  --args '{\"prompt\": \"abstract\"}' \\\n  --count 8 --randomize-seed --parallel 3 \\\n  --output-dir .\u002Foutputs\u002Fbatch\n```\n\n`-1` for `seed` (or omitting it with `--randomize-seed`) generates a fresh\nrandom seed per run.\n\n### Step 4: Present results\n\nThe scripts emit JSON to stdout describing every output file:\n\n```json\n{\n  \"status\": \"success\",\n  \"prompt_id\": \"abc-123\",\n  \"outputs\": [\n    {\"file\": \".\u002Foutputs\u002Fsdxl_00001_.png\", \"node_id\": \"9\",\n     \"type\": \"image\", \"filename\": \"sdxl_00001_.png\"}\n  ]\n}\n```\n\n## Decision Tree\n\n| User says | Tool | Command |\n|-----------|------|---------|\n| **Lifecycle (use comfy-cli)** | | |\n| \"install ComfyUI\" | comfy-cli | `bash scripts\u002Fcomfyui_setup.sh` |\n| \"start ComfyUI\" | comfy-cli | `comfy launch --background` |\n| \"stop ComfyUI\" | comfy-cli | `comfy stop` |\n| \"install X node\" | comfy-cli | `comfy node install \u003Cname>` |\n| \"download X model\" | comfy-cli | `comfy model download --url \u003Curl> --relative-path models\u002Fcheckpoints` |\n| \"list installed models\" | comfy-cli | `comfy model list` |\n| \"list installed nodes\" | comfy-cli | `comfy node show installed` |\n| **Execution (use scripts)** | | |\n| \"is everything ready?\" | script | `health_check.py` (optionally with `--workflow X --smoke-test`) |\n| \"what can I change in this workflow?\" | script | `extract_schema.py W.json` |\n| \"check if W's deps are met\" | script | `check_deps.py W.json` |\n| \"fix missing deps\" | script | `auto_fix_deps.py W.json` |\n| \"generate an image\" | script | `run_workflow.py --workflow W --args '{...}'` |\n| \"use this image\" (img2img) | script | `run_workflow.py --input-image image=.\u002Fx.png ...` |\n| \"8 variations with random seeds\" | script | `run_batch.py --count 8 --randomize-seed ...` |\n| \"show me live progress\" | script | `ws_monitor.py --prompt-id \u003Cid>` |\n| \"fetch the error from job X\" | script | `fetch_logs.py \u003Cprompt_id>` |\n| **Direct REST** | | |\n| \"what's in the queue?\" | REST | `curl http:\u002F\u002FHOST:8188\u002Fqueue` (local) or `--host https:\u002F\u002Fcloud.comfy.org` |\n| \"cancel that\" | REST | `curl -X POST http:\u002F\u002FHOST:8188\u002Finterrupt` |\n| \"free GPU memory\" | REST | `curl -X POST http:\u002F\u002FHOST:8188\u002Ffree` |\n\n## Setup & Onboarding\n\nWhen a user asks to set up ComfyUI, **the FIRST thing to do is ask whether\nthey want Comfy Cloud (hosted, zero install, API key) or Local (install\nComfyUI on their machine)**. Don't start running install commands or hardware\nchecks until they've answered.\n\n**Official docs:** https:\u002F\u002Fdocs.comfy.org\u002Finstallation\n**CLI docs:** https:\u002F\u002Fdocs.comfy.org\u002Fcomfy-cli\u002Fgetting-started\n**Cloud docs:** https:\u002F\u002Fdocs.comfy.org\u002Fget_started\u002Fcloud\n**Cloud API:** https:\u002F\u002Fdocs.comfy.org\u002Fdevelopment\u002Fcloud\u002Foverview\n\n### Step 0: Ask Local vs Cloud (ALWAYS FIRST)\n\nSuggested script:\n\n> \"Do you want to run ComfyUI locally on your machine, or use Comfy Cloud?\n>\n> - **Comfy Cloud** — hosted on RTX 6000 Pro GPUs, all common models pre-installed,\n>   zero setup. Requires an API key (paid subscription required to actually run\n>   workflows; free tier is read-only). Best if you don't have a capable GPU.\n> - **Local** — free, but your machine MUST meet the hardware requirements:\n>   - NVIDIA GPU with **≥6 GB VRAM** (≥8 GB for SDXL, ≥12 GB for Flux\u002Fvideo), OR\n>   - AMD GPU with ROCm support (Linux), OR\n>   - Apple Silicon Mac (M1+) with **≥16 GB unified memory** (≥32 GB recommended).\n>   - Intel Macs and machines with no GPU will NOT work — use Cloud instead.\n>\n> Which would you like?\"\n\nRouting:\n\n- **Cloud** → skip to **Path A**.\n- **Local** → run hardware check first, then pick a path from Paths B–E based on the verdict.\n- **Unsure** → run the hardware check and let the verdict decide.\n\n### Step 1: Verify Hardware (ONLY if user chose local)\n\n```bash\npython3 scripts\u002Fhardware_check.py --json\n# Optional: also probe `torch` for actual CUDA\u002FMPS:\npython3 scripts\u002Fhardware_check.py --json --check-pytorch\n```\n\n| Verdict    | Meaning                                                       | Action |\n|------------|---------------------------------------------------------------|--------|\n| `ok`       | ≥8 GB VRAM (discrete) OR ≥32 GB unified (Apple Silicon)       | Local install — use `comfy_cli_flag` from report |\n| `marginal` | SD1.5 works; SDXL tight; Flux\u002Fvideo unlikely                  | Local OK for light workflows, else **Path A (Cloud)** |\n| `cloud`    | No usable GPU, \u003C6 GB VRAM, \u003C16 GB Apple unified, Intel Mac, Rosetta Python | **Switch to Cloud** unless user explicitly forces local |\n\nThe script also surfaces `wsl: true` (WSL2 with NVIDIA passthrough) and\n`rosetta: true` (x86_64 Python on Apple Silicon — must reinstall as ARM64).\n\nIf verdict is `cloud` but the user wants local, do not proceed silently.\nShow the `notes` array verbatim and ask whether they want to (a) switch to\nCloud or (b) force a local install (will OOM or be unusably slow on modern models).\n\n### Choosing an Installation Path\n\nUse the hardware check first. The table below is the fallback for when the\nuser has already told you their hardware:\n\n| Situation | Recommended Path |\n|-----------|------------------|\n| `verdict: cloud` from hardware check | **Path A: Comfy Cloud** |\n| No GPU \u002F want to try without commitment | **Path A: Comfy Cloud** |\n| Windows + NVIDIA + non-technical | **Path B: ComfyUI Desktop** |\n| Windows + NVIDIA + technical | **Path C: Portable** or **Path D: comfy-cli** |\n| Linux + any GPU | **Path D: comfy-cli** (easiest) |\n| macOS + Apple Silicon | **Path B: Desktop** or **Path D: comfy-cli** |\n| Headless \u002F server \u002F CI \u002F agents | **Path D: comfy-cli** |\n\nFor the fully automated path (hardware check → install → launch → verify):\n\n```bash\nbash scripts\u002Fcomfyui_setup.sh\n# Or with overrides:\nbash scripts\u002Fcomfyui_setup.sh --m-series --port=8190 --workspace=\u002Fdata\u002Fcomfy\n```\n\nIt runs `hardware_check.py` internally, refuses to install locally when the\nverdict is `cloud` (unless `--force-cloud-override`), picks the right\n`comfy-cli` flag, and prefers `pipx`\u002F`uvx` over global `pip` to avoid polluting\nsystem Python.\n\n---\n\n### Path A: Comfy Cloud (No Local Install)\n\nFor users without a capable GPU or who want zero setup. Hosted on RTX 6000 Pro.\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Fget_started\u002Fcloud\n\n1. Sign up at https:\u002F\u002Fcomfy.org\u002Fcloud\n2. Generate an API key at https:\u002F\u002Fplatform.comfy.org\u002Flogin\n3. Set the key:\n   ```bash\n   export COMFY_CLOUD_API_KEY=\"comfyui-xxxxxxxxxxxx\"\n   ```\n4. Run workflows:\n   ```bash\n   python3 scripts\u002Frun_workflow.py \\\n     --workflow workflows\u002Fflux_dev_txt2img.json \\\n     --args '{\"prompt\": \"...\"}' \\\n     --host https:\u002F\u002Fcloud.comfy.org \\\n     --output-dir .\u002Foutputs\n   ```\n\n**Pricing:** https:\u002F\u002Fwww.comfy.org\u002Fcloud\u002Fpricing\n**Concurrent jobs:** Free\u002FStandard 1, Creator 3, Pro 5. Free tier\n**cannot run workflows via API** — only browse models. Paid subscription\nrequired for `\u002Fapi\u002Fprompt`, `\u002Fapi\u002Fupload\u002F*`, `\u002Fapi\u002Fview`, etc.\n\n---\n\n### Path B: ComfyUI Desktop (Windows \u002F macOS)\n\nOne-click installer for non-technical users. Currently Beta.\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Finstallation\u002Fdesktop\n- **Windows (NVIDIA):** https:\u002F\u002Fdownload.comfy.org\u002Fwindows\u002Fnsis\u002Fx64\n- **macOS (Apple Silicon):** https:\u002F\u002Fcomfy.org\n\nLinux is **not supported** for Desktop — use Path D.\n\n---\n\n### Path C: ComfyUI Portable (Windows Only)\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Finstallation\u002Fcomfyui_portable_windows\n\nDownload from https:\u002F\u002Fgithub.com\u002Fcomfyanonymous\u002FComfyUI\u002Freleases, extract,\nrun `run_nvidia_gpu.bat`. Update via `update\u002Fupdate_comfyui_stable.bat`.\n\n---\n\n### Path D: comfy-cli (All Platforms — Recommended for Agents)\n\nThe official CLI is the best path for headless\u002Fautomated setups.\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Fcomfy-cli\u002Fgetting-started\n\n#### Install comfy-cli\n\n```bash\n# Recommended:\npipx install comfy-cli\n# Or use uvx without installing:\nuvx --from comfy-cli comfy --help\n# Or (if pipx\u002Fuvx unavailable):\npip install --user comfy-cli\n```\n\nDisable analytics non-interactively:\n```bash\ncomfy --skip-prompt tracking disable\n```\n\n#### Install ComfyUI\n\n```bash\ncomfy --skip-prompt install --nvidia              # NVIDIA (CUDA)\ncomfy --skip-prompt install --amd                 # AMD (ROCm, Linux)\ncomfy --skip-prompt install --m-series            # Apple Silicon (MPS)\ncomfy --skip-prompt install --cpu                 # CPU only (slow)\ncomfy --skip-prompt install --nvidia --fast-deps  # uv-based dep resolution\n```\n\nDefault location: `~\u002Fcomfy\u002FComfyUI` (Linux), `~\u002FDocuments\u002Fcomfy\u002FComfyUI`\n(macOS\u002FWin). Override with `comfy --workspace \u002Fcustom\u002Fpath install`.\n\n#### Launch \u002F verify\n\n```bash\ncomfy launch --background                       # background daemon on :8188\ncomfy launch -- --listen 0.0.0.0 --port 8190    # LAN-accessible custom port\ncurl -s http:\u002F\u002F127.0.0.1:8188\u002Fsystem_stats      # health check\n```\n\n---\n\n### Path E: Manual Install (Advanced \u002F Unsupported Hardware)\n\nFor Ascend NPU, Cambricon MLU, Intel Arc, or other unsupported hardware.\n\n**Docs:** https:\u002F\u002Fdocs.comfy.org\u002Finstallation\u002Fmanual_install\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002Fcomfyanonymous\u002FComfyUI.git\ncd ComfyUI\npip install torch torchvision torchaudio --extra-index-url https:\u002F\u002Fdownload.pytorch.org\u002Fwhl\u002Fcu130\npip install -r requirements.txt\npython main.py\n```\n\n---\n\n### Post-Install: Download Models\n\n```bash\n# SDXL (general purpose, ~6.5 GB)\ncomfy model download \\\n  --url \"https:\u002F\u002Fhuggingface.co\u002Fstabilityai\u002Fstable-diffusion-xl-base-1.0\u002Fresolve\u002Fmain\u002Fsd_xl_base_1.0.safetensors\" \\\n  --relative-path models\u002Fcheckpoints\n\n# SD 1.5 (lighter, ~4 GB, good for 6 GB cards)\ncomfy model download \\\n  --url \"https:\u002F\u002Fhuggingface.co\u002Fstable-diffusion-v1-5\u002Fstable-diffusion-v1-5\u002Fresolve\u002Fmain\u002Fv1-5-pruned-emaonly.safetensors\" \\\n  --relative-path models\u002Fcheckpoints\n\n# Flux Dev fp8 (smaller variant, ~12 GB)\ncomfy model download \\\n  --url \"https:\u002F\u002Fhuggingface.co\u002FComfy-Org\u002Fflux1-dev\u002Fresolve\u002Fmain\u002Fflux1-dev-fp8.safetensors\" \\\n  --relative-path models\u002Fcheckpoints\n\n# CivitAI (set token first):\ncomfy model download \\\n  --url \"https:\u002F\u002Fcivitai.com\u002Fapi\u002Fdownload\u002Fmodels\u002F128713\" \\\n  --relative-path models\u002Fcheckpoints \\\n  --set-civitai-api-token \"YOUR_TOKEN\"\n```\n\nList installed: `comfy model list`.\n\n### Post-Install: Install Custom Nodes\n\n```bash\ncomfy node install comfyui-impact-pack             # popular utility pack\ncomfy node install comfyui-animatediff-evolved     # video generation\ncomfy node install comfyui-controlnet-aux          # ControlNet preprocessors\ncomfy node install comfyui-essentials              # common helpers\ncomfy node update all\ncomfy node install-deps --workflow=workflow.json   # install everything a workflow needs\n```\n\n### Post-Install: Verify\n\n```bash\npython3 scripts\u002Fhealth_check.py\n# → comfy_cli on PATH? server reachable? checkpoints? smoke test?\n\npython3 scripts\u002Fcheck_deps.py my_workflow.json\n# → are this workflow's nodes\u002Fmodels\u002Fembeddings installed?\n\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflows\u002Fsd15_txt2img.json \\\n  --args '{\"prompt\": \"test\", \"steps\": 4}' \\\n  --output-dir .\u002Ftest-outputs\n```\n\n## Image Upload (img2img \u002F Inpainting)\n\nThe simplest way is to use `--input-image` with `run_workflow.py`:\n\n```bash\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflows\u002Fsdxl_img2img.json \\\n  --input-image image=.\u002Fphoto.png \\\n  --args '{\"prompt\": \"make it cyberpunk\", \"denoise\": 0.6}'\n```\n\nThe flag uploads `photo.png`, then injects its server-side filename into\nwhatever schema parameter is named `image`. For inpainting, pass both:\n\n```bash\npython3 scripts\u002Frun_workflow.py \\\n  --workflow workflows\u002Fsdxl_inpaint.json \\\n  --input-image image=.\u002Fphoto.png \\\n  --input-image mask_image=.\u002Fmask.png \\\n  --args '{\"prompt\": \"fill with flowers\"}'\n```\n\nManual upload via REST:\n```bash\ncurl -X POST \"http:\u002F\u002F127.0.0.1:8188\u002Fupload\u002Fimage\" \\\n  -F \"image=@photo.png\" -F \"type=input\" -F \"overwrite=true\"\n# Returns: {\"name\": \"photo.png\", \"subfolder\": \"\", \"type\": \"input\"}\n\n# Cloud equivalent:\ncurl -X POST \"https:\u002F\u002Fcloud.comfy.org\u002Fapi\u002Fupload\u002Fimage\" \\\n  -H \"X-API-Key: $COMFY_CLOUD_API_KEY\" \\\n  -F \"image=@photo.png\" -F \"type=input\" -F \"overwrite=true\"\n```\n\n## Cloud Specifics\n\n- **Base URL:** `https:\u002F\u002Fcloud.comfy.org`\n- **Auth:** `X-API-Key` header (or `?token=KEY` for WebSocket)\n- **API key:** set `$COMFY_CLOUD_API_KEY` once and the scripts pick it up automatically\n- **Output download:** `\u002Fapi\u002Fview` returns a 302 to a signed URL; the scripts\n  follow it and strip `X-API-Key` before fetching from the storage backend\n  (don't leak the API key to S3\u002FCloudFront).\n- **Endpoint differences from local ComfyUI:**\n  - `\u002Fapi\u002Fobject_info`, `\u002Fapi\u002Fqueue`, `\u002Fapi\u002Fuserdata` — **403 on free tier**;\n    paid only.\n  - `\u002Fhistory` is renamed to `\u002Fhistory_v2` on cloud (the scripts route\n    automatically).\n  - `\u002Fmodels\u002F\u003Cfolder>` is renamed to `\u002Fexperiment\u002Fmodels\u002F\u003Cfolder>` on cloud\n    (the scripts route automatically).\n  - `clientId` in WebSocket is currently ignored — all connections for a\n    user receive the same broadcast. Filter by `prompt_id` client-side.\n  - `subfolder` is accepted on uploads but ignored — cloud has a flat namespace.\n- **Concurrent jobs:** Free\u002FStandard: 1, Creator: 3, Pro: 5. Extras queue\n  automatically. Use `run_batch.py --parallel N` to saturate your tier.\n\n## Queue & System Management\n\n```bash\n# Local\ncurl -s http:\u002F\u002F127.0.0.1:8188\u002Fqueue | python3 -m json.tool\ncurl -X POST http:\u002F\u002F127.0.0.1:8188\u002Fqueue -d '{\"clear\": true}'    # cancel pending\ncurl -X POST http:\u002F\u002F127.0.0.1:8188\u002Finterrupt                      # cancel running\ncurl -X POST http:\u002F\u002F127.0.0.1:8188\u002Ffree \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"unload_models\": true, \"free_memory\": true}'\n\n# Cloud — same paths under \u002Fapi\u002F, plus:\npython3 scripts\u002Ffetch_logs.py --tail-queue --host https:\u002F\u002Fcloud.comfy.org\n```\n\n## Pitfalls\n\n1. **API format required** — every script and the `\u002Fapi\u002Fprompt` endpoint expect\n   API-format workflow JSON. The scripts detect editor format (top-level\n   `nodes` and `links` arrays) and tell you to re-export via\n   \"Workflow → Export (API)\" (newer UI) or \"Save (API Format)\" (older UI).\n\n2. **Server must be running** — all execution requires a live server.\n   `comfy launch --background` starts one. Verify with\n   `curl http:\u002F\u002F127.0.0.1:8188\u002Fsystem_stats`.\n\n3. **Model names are exact** — case-sensitive, includes file extension.\n   `check_deps.py` does fuzzy matching (with\u002Fwithout extension and folder\n   prefix), but the workflow itself must use the canonical name. Use\n   `comfy model list` to discover what's installed.\n\n4. **Missing custom nodes** — \"class_type not found\" means a required node\n   isn't installed. `check_deps.py` reports which package to install;\n   `auto_fix_deps.py` runs the install for you.\n\n5. **Working directory** — `comfy-cli` auto-detects the ComfyUI workspace.\n   If commands fail with \"no workspace found\", use\n   `comfy --workspace \u002Fpath\u002Fto\u002FComfyUI \u003Ccommand>` or\n   `comfy set-default \u002Fpath\u002Fto\u002FComfyUI`.\n\n6. **Cloud free-tier API limits** — `\u002Fapi\u002Fprompt`, `\u002Fapi\u002Fview`, `\u002Fapi\u002Fupload\u002F*`,\n   `\u002Fapi\u002Fobject_info` all return 403 on free accounts. `health_check.py` and\n   `check_deps.py` handle this gracefully and surface a clear message.\n\n7. **Timeout for video\u002Faudio workflows** — auto-detected when an output node\n   is `VHS_VideoCombine`, `SaveVideo`, etc.; the default jumps from 300 s to\n   900 s. Override explicitly with `--timeout 1800`.\n\n8. **Path traversal in output filenames** — server-supplied filenames are\n   passed through `safe_path_join` to refuse anything escaping `--output-dir`.\n   Keep this protection on — workflows with custom save nodes can produce\n   arbitrary paths.\n\n9. **Workflow JSON is arbitrary code** — custom nodes run Python, so\n   submitting an unknown workflow has the same trust profile as `eval`.\n   Inspect workflows from untrusted sources before running.\n\n10. **Auto-randomized seed** — pass `seed: -1` in `--args` (or use\n    `--randomize-seed` and omit the seed) to get a fresh seed per run.\n    The actual seed is logged to stderr.\n\n11. **`tracking` prompt** — first run of `comfy` may prompt for analytics.\n    Use `comfy --skip-prompt tracking disable` to skip non-interactively.\n    `comfyui_setup.sh` does this for you.\n\n## Verification Checklist\n\nUse `python3 scripts\u002Fhealth_check.py` to run the whole list at once. Manual:\n\n- [ ] `hardware_check.py` verdict is `ok` OR the user explicitly chose Comfy Cloud\n- [ ] `comfy --version` works (or `uvx --from comfy-cli comfy --help`)\n- [ ] `curl http:\u002F\u002FHOST:PORT\u002Fsystem_stats` returns JSON\n- [ ] `comfy model list` shows at least one checkpoint (local) OR\n      `\u002Fapi\u002Fexperiment\u002Fmodels\u002Fcheckpoints` returns models (cloud)\n- [ ] Workflow JSON is in API format\n- [ ] `check_deps.py` reports `is_ready: true` (or only `node_check_skipped`\n      on cloud free tier)\n- [ ] Test run with a small workflow completes; outputs land in `--output-dir`\n",{"name":5,"setup":24,"author":26,"license":15,"version":30,"metadata":31,"platforms":46,"description":7,"compatibility":50,"prerequisites":51},{"help":25},"Run scripts\u002Fhardware_check.py FIRST to decide local vs Comfy Cloud; then scripts\u002Fcomfyui_setup.sh auto-installs locally (or use Cloud API key for platform.comfy.org).",[27,28,29],"kshitijk4poor","alt-glitch","purzbeats","5.1.0",{"hermes":32},{"tags":33,"category":40,"related_skills":43},[5,34,35,36,37,38,39,40,41,42],"image-generation","stable-diffusion","flux","sd3","wan-video","hunyuan-video","creative","generative-ai","video-generation",[44,45],"stable-diffusion-image-generation","image_gen",[47,48,49],"macos","linux","windows","Requires ComfyUI (local, Comfy Desktop, or Comfy Cloud) and comfy-cli (auto-installed via pipx\u002Fuvx by the setup script).",{"commands":52},[53],"python3","main","eac438ea9216a7cb8a570cab803eb01eeafc6793","upstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m","https:\u002F\u002Fid-preview--d5bef06d-6c9e-4563-a3d4-a855b9d61610.lovable.app\u002F","Eu sou Antonio B. B. Ndombe \r\nEu sou um Pesquisador e desenvolvedor de sistema autonomia cognitivos, fundador de sistemas ODCs ( organismos digital conscientes)","2026-11-04 06:53:39",{"source":8,"skillName":5,"commands":63,"prompt":73,"zipAvailable":74},[64,67,70],{"tool":65,"command":66},"npx","npx skills add GrupoANDevelopment-m\u002Falberto-ai --skill comfyui",{"tool":68,"command":69},"bunx","bunx skills add GrupoANDevelopment-m\u002Falberto-ai --skill comfyui",{"tool":71,"command":72},"pnpm","pnpm dlx skills add GrupoANDevelopment-m\u002Falberto-ai --skill comfyui","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002FSKILL.md",true,[76,82,86,90,94,100,104,108,112,116,120,124,128,132,136,140,144,150,154,158,162,166,170,174,178,182,186,190,194,198,202,206],{"path":77,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":80,"sourceHtmlUrl":81},"references\u002Fofficial-cli.md","REFERENCE","参考文档",7531,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Freferences\u002Fofficial-cli.md",{"path":83,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":84,"sourceHtmlUrl":85},"references\u002Frest-api.md",10363,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Freferences\u002Frest-api.md",{"path":87,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":88,"sourceHtmlUrl":89},"references\u002Ftemplate-integrity.md",8821,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Freferences\u002Ftemplate-integrity.md",{"path":91,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":92,"sourceHtmlUrl":93},"references\u002Fworkflow-format.md",8025,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Freferences\u002Fworkflow-format.md",{"path":95,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":98,"sourceHtmlUrl":99},"scripts\u002F_common.py","SCRIPT","脚本",33742,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002F_common.py",{"path":101,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":102,"sourceHtmlUrl":103},"scripts\u002Fauto_fix_deps.py",8279,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Fauto_fix_deps.py",{"path":105,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":106,"sourceHtmlUrl":107},"scripts\u002Fcheck_deps.py",17164,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Fcheck_deps.py",{"path":109,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":110,"sourceHtmlUrl":111},"scripts\u002Fcomfyui_setup.sh",10447,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Fcomfyui_setup.sh",{"path":113,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":114,"sourceHtmlUrl":115},"scripts\u002Fextract_schema.py",11289,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Fextract_schema.py",{"path":117,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":118,"sourceHtmlUrl":119},"scripts\u002Ffetch_logs.py",5690,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Ffetch_logs.py",{"path":121,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":122,"sourceHtmlUrl":123},"scripts\u002Fhardware_check.py",17833,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Fhardware_check.py",{"path":125,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":126,"sourceHtmlUrl":127},"scripts\u002Fhealth_check.py",8188,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Fhealth_check.py",{"path":129,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":130,"sourceHtmlUrl":131},"scripts\u002Frun_batch.py",9080,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Frun_batch.py",{"path":133,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":134,"sourceHtmlUrl":135},"scripts\u002Frun_workflow.py",31591,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Frun_workflow.py",{"path":137,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":138,"sourceHtmlUrl":139},"scripts\u002Fws_monitor.py",10512,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fscripts\u002Fws_monitor.py",{"path":141,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":142,"sourceHtmlUrl":143},"tests\u002Fconftest.py",1480,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Ftests\u002Fconftest.py",{"path":145,"resourceType":146,"resourceTypeDesc":147,"sizeBytes":148,"sourceHtmlUrl":149},"tests\u002Fpytest.ini","OTHER","其他",132,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Ftests\u002Fpytest.ini",{"path":151,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":152,"sourceHtmlUrl":153},"tests\u002FREADME.md",2124,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Ftests\u002FREADME.md",{"path":155,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":156,"sourceHtmlUrl":157},"tests\u002Ftest_check_deps.py",2362,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Ftests\u002Ftest_check_deps.py",{"path":159,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":160,"sourceHtmlUrl":161},"tests\u002Ftest_cloud_integration.py",4324,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Ftests\u002Ftest_cloud_integration.py",{"path":163,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":164,"sourceHtmlUrl":165},"tests\u002Ftest_common.py",16337,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Ftests\u002Ftest_common.py",{"path":167,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":168,"sourceHtmlUrl":169},"tests\u002Ftest_extract_schema.py",7972,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Ftests\u002Ftest_extract_schema.py",{"path":171,"resourceType":96,"resourceTypeDesc":97,"sizeBytes":172,"sourceHtmlUrl":173},"tests\u002Ftest_run_workflow.py",8708,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Ftests\u002Ftest_run_workflow.py",{"path":175,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":176,"sourceHtmlUrl":177},"workflows\u002Fanimatediff_video.json",2071,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002Fanimatediff_video.json",{"path":179,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":180,"sourceHtmlUrl":181},"workflows\u002Fflux_dev_txt2img.json",2232,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002Fflux_dev_txt2img.json",{"path":183,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":184,"sourceHtmlUrl":185},"workflows\u002FREADME.md",3932,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002FREADME.md",{"path":187,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":188,"sourceHtmlUrl":189},"workflows\u002Fsd15_txt2img.json",1474,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002Fsd15_txt2img.json",{"path":191,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":192,"sourceHtmlUrl":193},"workflows\u002Fsdxl_img2img.json",1652,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002Fsdxl_img2img.json",{"path":195,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":196,"sourceHtmlUrl":197},"workflows\u002Fsdxl_inpaint.json",1873,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002Fsdxl_inpaint.json",{"path":199,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":200,"sourceHtmlUrl":201},"workflows\u002Fsdxl_txt2img.json",1449,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002Fsdxl_txt2img.json",{"path":203,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":204,"sourceHtmlUrl":205},"workflows\u002Fupscale_4x.json",864,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002Fupscale_4x.json",{"path":207,"resourceType":78,"resourceTypeDesc":79,"sizeBytes":208,"sourceHtmlUrl":209},"workflows\u002Fwan_video_t2v.json",2189,"https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fhermes\u002Fskills\u002Fcreative\u002Fcomfyui\u002Fworkflows\u002Fwan_video_t2v.json",[211,219,226,233,240,247],{"skillNo":212,"name":213,"nameZh":6,"description":214,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":215,"licenseFileUrl":6,"sourceHtmlUrl":216,"categories":217,"keywords":218,"publishedAt":19},"01M7DASC6FSCJP49JZBSK0NY1Z","nemoclaw-user-reference","Describes the NemoClaw integration layer and blueprint architecture and how they orchestrate compatible agent sandboxes. Use when looking up architecture, agent integration, plugin structure, or blueprint design. Trigger keywords - nemoclaw architecture, nemoclaw agent architecture, nemoclaw plugin blueprint structure, nemoclaw vs openshell, which cli, nemoclaw cli, openshell cli, sandbox commands, nemoclaw cli commands, nemoclaw command reference, nemoclaw network policy, sandbox egress control operator approval, nemoclaw troubleshooting, nemoclaw debug sandbox issues.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fnemoclaw\u002Fskills\u002Fnemoclaw-user-reference\u002FSKILL.md",[],[],{"skillNo":220,"name":221,"nameZh":6,"description":222,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":215,"licenseFileUrl":6,"sourceHtmlUrl":223,"categories":224,"keywords":225,"publishedAt":19},"01M7DAS5S79VK33XFGVAM2TFWF","nemoclaw-user-overview","Explains how OpenClaw, OpenShell, and NemoClaw form the ecosystem, NemoClaw's position in the stack, what NemoClaw adds beyond the community sandbox, and when to prefer NemoClaw versus integrating OpenShell and OpenClaw directly. Use when users ask about the relationship between OpenClaw, OpenShell, and NemoClaw, or when to use NemoClaw versus OpenShell. Trigger keywords - nemoclaw ecosystem, openclaw openshell, nemoclaw vs openshell, sandboxed openclaw, how nemoclaw works, nemoclaw sandbox lifecycle blueprint, nemoclaw overview, openclaw always-on assistants, nvidia openshell, nvidia nemotron, nemoclaw release notes, nemoclaw changelog.","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fnemoclaw\u002Fskills\u002Fnemoclaw-user-overview\u002FSKILL.md",[],[],{"skillNo":227,"name":228,"nameZh":6,"description":229,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":215,"licenseFileUrl":6,"sourceHtmlUrl":230,"categories":231,"keywords":232,"publishedAt":19},"01M7DAS0PQEMET5MNGZ1ABSZVS","nemoclaw-user-monitor-sandbox","Inspects sandbox health, traces agent behavior, and diagnoses problems. Use when monitoring a running sandbox, debugging agent issues, or checking sandbox logs. Trigger keywords - monitor nemoclaw sandbox, debug nemoclaw agent issues.","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fnemoclaw\u002Fskills\u002Fnemoclaw-user-monitor-sandbox\u002FSKILL.md",[],[],{"skillNo":234,"name":235,"nameZh":6,"description":236,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":215,"licenseFileUrl":6,"sourceHtmlUrl":237,"categories":238,"keywords":239,"publishedAt":19},"01M7DARQYGF7E9A69VH6S3CGN9","nemoclaw-user-manage-sandboxes","Explains operational tasks after the quickstart: listing sandboxes, status and health checks, logs, diagnostics, port forwards, multiple sandboxes, credential reset, rebuilds, network presets, upgrades, and uninstall. Trigger keywords - manage nemoclaw sandboxes, nemoclaw status, nemoclaw list, nemoclaw dashboard port, nemoclaw rebuild, nemoclaw upgrade sandboxes, nemoclaw uninstall, sandbox mutability, sandbox runtime configuration, sandbox rebuild, nemoclaw backup, nemoclaw restore, workspace backup, openshell sandbox download upload, nemoclaw messaging channels, nemoclaw telegram, nemoclaw discord, nemoclaw slack, nemoclaw wechat, nemoclaw whatsapp, openshell channel messaging, nemoclaw workspace files, soul.md, user.md, identity.md, agents.md, sandbox persistence.","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fnemoclaw\u002Fskills\u002Fnemoclaw-user-manage-sandboxes\u002FSKILL.md",[],[],{"skillNo":241,"name":242,"nameZh":6,"description":243,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":215,"licenseFileUrl":6,"sourceHtmlUrl":244,"categories":245,"keywords":246,"publishedAt":19},"01M7DARH8BRTG51TRCAB7YWDJ4","nemoclaw-user-manage-policy","Adds, removes, or modifies allowed endpoints in the sandbox policy. Use when customizing network policy, changing egress rules, or configuring sandbox endpoint access. Trigger keywords - customize nemoclaw network policy, sandbox egress policy configuration, nemoclaw integration policy examples, post-install policy setup, openshell approval workflow, policy preset, nemoclaw approve network requests, sandbox egress approval tui.","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fnemoclaw\u002Fskills\u002Fnemoclaw-user-manage-policy\u002FSKILL.md",[],[],{"skillNo":248,"name":249,"nameZh":6,"description":250,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":215,"licenseFileUrl":6,"sourceHtmlUrl":251,"categories":252,"keywords":253,"publishedAt":19},"01M7DARAT62M2VR37590VCZ5NS","nemoclaw-user-get-started","Installs NemoClaw, launches a sandbox, and runs the first agent prompt. Use when onboarding, installing, or launching a NemoClaw sandbox for the first time. Trigger keywords - nemoclaw quickstart, install nemoclaw openclaw sandbox, nemohermes quickstart, hermes agent nemoclaw, run hermes openshell sandbox, nemoclaw prerequisites, nemoclaw supported platforms, nemoclaw hardware software, nemoclaw windows wsl2 setup, nemoclaw install windows docker desktop.","https:\u002F\u002Fgithub.com\u002FGrupoANDevelopment-m\u002Falberto-ai\u002Fblob\u002Fmain\u002Fupstream\u002Fnemoclaw\u002Fskills\u002Fnemoclaw-user-get-started\u002FSKILL.md",[],[]]