メニュー
Use only when the selected HOI4 portrait provider is a configured loopback ComfyUI installation.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Use only when the selected HOI4 portrait provider is the Comfy Cloud MCP route.
Use only when the selected HOI4 portrait provider is an existing RunPod ComfyUI workspace.
Use when creating, sourcing, processing, converting, organizing, wiring, or documenting visual assets for a Hearts of Iron IV mod.
Use for the provider-neutral HOI4 portrait source, prompt, archive, validation, and runtime handoff contract.
Use when coordinating custom Codex subagents for HOI4 mod implementation, asset production, text and audio research, audits, active small patches, planning handoffs, or documentation work.
Use when implementing, updating, auditing, or documenting Hearts of Iron IV events and event chains, including news and report events, on_action hooks, localisation, pictures, registration hooks, and validation.
| name | hoi4-comfyui-local |
| description | Use only when the selected HOI4 portrait provider is a configured loopback ComfyUI installation. |
Read hoi4-portrait-production first. This is the only provider-specific
portrait route that may be used for this project. Do not use another provider
skill or scan the computer for an unapproved installation.
Use only an explicitly configured or verified common installation root and a
loopback server at 127.0.0.1, localhost, or ::1. Report the root, server
URL, exact workflow commit, model presence, hardware result, and provider
status.
Before offering a new installation, inspect hardware and refuse an infeasible route unless the user explicitly overrides the gate. Use the exact upstream scripts after review:
.\scripts\install_windows.ps1 -ComfyUIRoot "<COMFYUI_ROOT>"
python scripts/install_workflows.py --comfyui-root "<COMFYUI_ROOT>"
python scripts/download_models.py --comfyui-root "<COMFYUI_ROOT>"
The current manifest installs eight pinned model files totaling about 19.42 GB
decimal. A 24 GB GPU is the practical target; 18 GB may work with aggressive
offloading, and 16 GB is limited to slower reduced-resolution tests. Accept
the gated FLUX.2 Klein agreement before downloading the base model. The
current source and processing graphs also require the
adaptive_portrait_crop custom node and the MediaPipe/YuNet detector files.
Use the canonical model manifest and require Hugging Face authorization where the manifest requires it. Submit through the loopback REST/WebSocket interface, poll the exact job/history record, retrieve both output sizes, and keep server or filesystem paths out of runtime portrait references.
Missing hardware, authorization, models, workflow installation, or a running server is an honest incomplete state. It is not successful portrait production.