메뉴
Manage local AI runtimes (Ollama, ComfyUI, Stable Diffusion WebUI) on the ASUS GX10 workstation. Use when starting, stopping, or checking status of local LLMs and image generators.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Professional orchestration of the ACG Content Factory and "Brainless" Gumroad distribution. Use when creating premium character packs, managing Gumroad listings, or bypassing API limitations.
Knowledge about the round-robin babysitter script, /run slash commands, and how Aider should execute local scripts rather than just describing them. Use when orchestrating Aider sessions or teaching Aider to self-execute.
Manage the Children's Book Factory pipeline. Use when generating story drafts, running peer reviews, assembling PDFs, or orchestrating the automated book production workflow.
Generate images locally using ComfyUI (port 8188) on the ASUS GX10. Use when creating illustrations for books, character art, ACG content, or any image generation task. Includes batch generation, theme packs, and upload to catbox.moe.
Specialized knowledge of the ASUS GX10 workstation architecture, environment constraints, development workflows, and business operations. Use when troubleshooting environment issues or architecting new integrations.
| name | gx10-orchestrator |
| description | Manage local AI runtimes (Ollama, ComfyUI, Stable Diffusion WebUI) on the ASUS GX10 workstation. Use when starting, stopping, or checking status of local LLMs and image generators. |
| Item | Value |
|---|---|
| SoC | NVIDIA Grace Blackwell GB10 (ARM aarch64) |
| RAM | 128 GB LPDDR5x Coherent Unified Memory (CPU+GPU shared) |
| OS | Ubuntu 24.04.4 LTS, CUDA 13.0 |
| Storage | 916 GB NVMe, ~598 GB free |
| Tailscale | 100.108.147.12 |
| Domains | ai.hjs.space → ComfyUI :8188 |
| Item | Value |
|---|---|
| Binary | /home/hjs/bin/ollama |
| API | http://127.0.0.1:11434 |
| Restart | systemctl --user restart ollama |
| Model | Size | Role |
|---|---|---|
qwen3.5-queen-27b | 28 GB | Primary (drafts, review, char gen) |
gemma-4-31b-uncensored | 32 GB | Creative fallback, uncensored |
golddiamondgold-70b | 42 GB | High-quality heavy fallback |
glm-4-9b-abliterated | 10 GB | Fast lightweight tasks |
hermes3 | 4.7 GB | Tool-use / function calling |
nemotron-mini | 2.7 GB | Ultra-fast triage |
import urllib.request, json
def ollama_generate(model: str, prompt: str) -> str:
payload = json.dumps({"model": model, "prompt": prompt, "stream": False}).encode()
req = urllib.request.Request("http://127.0.0.1:11434/api/generate",
data=payload, headers={"Content-Type": "application/json"})
with urllib.request.urlopen(req, timeout=120) as r:
return json.loads(r.read())["response"].strip()
| Item | Value |
|---|---|
| Location | /home/hjs/ComfyUI |
| Venv | /home/hjs/comfyui_env |
| Port | 8188 |
| Model | animagine-xl-3.1.safetensors (symlink) |
| Speed | ~67s/image |
# Start
source /home/hjs/comfyui_env/bin/activate
nohup python3 /home/hjs/ComfyUI/main.py --listen 127.0.0.1 --port 8188 > /tmp/comfyui.log 2>&1 &
# Health check
curl -s http://127.0.0.1:8188/system_stats | python3 -c "import sys,json; d=json.load(sys.stdin); print('OK v'+d['system']['comfyui_version'])"
/home/hjs/stable-diffusion-webui | Port: 7860cd ~/stable-diffusion-webui && ./webui.sh# Symlink SD WebUI checkpoints → ComfyUI (already done for animagine)
ln -sf ~/stable-diffusion-webui/models/Stable-diffusion/animagine-xl-3.1.safetensors \
~/ComfyUI/models/checkpoints/animagine-xl-3.1.safetensors
qwen3.5-queen-27b (local primary)gemma-4-31b-uncensored (local fallback)golddiamondgold-70b (heavy local last resort)