AI PC Screen Doctor

Use this skill when the user needs a structured diagnosis for a screen error rather than a generic chat response.

Inputs

• image_path: optional local screenshot path
• ocr_text: optional copied error text or OCR text
• context: optional environment, system version, recent actions, and reproduction notes

Output

Return a JSON diagnosis with these fields:

{
  "engine": "openclaw-local-rules",
  "id": "python_missing_module",
  "title": "Python 依赖缺失或解释器环境不一致",
  "category": "Python dependency",
  "severity": "medium",
  "confidence": 0.88,
  "summary": "结构化诊断摘要",
  "signals": [],
  "probable_causes": [],
  "immediate_actions": [],
  "commands": [],
  "verification": [],
  "escalation": []
}

Workflow

1. Read ocr_text and context.
2. If image_path exists, treat it as supporting context; do not require image understanding to proceed.
3. Match the input against known fault patterns.
4. Produce a low-risk, executable diagnosis.
5. Prefer concrete commands and verification steps over generic advice.

Covered cases

• Python dependency or interpreter mismatch
• Node install or permission failures
• GPU / NPU memory exhaustion
• Display driver failures, black screen, blue screen
• Generic screen failures with missing context

Local execution

cd ai_pc_screen_doctor_skill\skill
python skill.py --text "ModuleNotFoundError: No module named openvino_genai" --context "Windows 11, venv changed"

With an image path:

python skill.py --image ".\sample.png" --text "VIDEO_TDR_FAILURE" --context "外接显示器后偶发黑屏蓝屏"

Runtime notes

• Python >= 3.9
• No external API is required for the minimal skill
• Optional environment variables for the full project:
  • DASHSCOPE_API_KEY
  • DEEPSEEK_API_KEY

Files

• manifest.json: skill metadata
• skill.py: CLI entry and local rules engine