skills/automatic1111/_scores.json

{
  "version": "1.0.0",
  "skillHash": "sha256:612ef2b5dde4578c2b288aa474643da0ef7036e3cdf51719dfec7f97fcb5c15d",
  "scoredAt": "2026-05-13T14:20:00.716Z",
  "backend": "ollama",
  "model": "gpt-oss:20b",
  "quality": {
    "score": 87,
    "dimensions": {
      "clarity": "PASS",
      "completeness": "WEAK",
      "conciseness": "PASS",
      "actionability": "PASS",
      "crossPlatform": "WEAK",
      "examples": "PASS"
    },
    "issues": [
      {
        "severity": "MEDIUM",
        "category": "completeness",
        "detail": "Missing error handling and edge case coverage."
      },
      {
        "severity": "MEDIUM",
        "category": "crossPlatform",
        "detail": "Minor platform assumptions (bash scripts) limit cross-platform compatibility."
      }
    ]
  },
  "security": {
    "verdict": "SUSPICIOUS",
    "issues": [
      {
        "severity": "MEDIUM",
        "category": "checksumMissing",
        "detail": "Missing checksum verification for downloaded model file."
      }
    ]
  },
  "impact": {
    "multiplier": 4,
    "baselineAvg": 25,
    "treatmentAvg": 100,
    "scenarios": [
      {
        "name": "lora-image-generation",
        "baseline": 0,
        "treatment": 100,
        "rationale": "Response A fully satisfies the rubric, while Response B omits the required prompt tag, negative_prompt, and uses an unsupported payload structure."
      },
      {
        "name": "inpainting-with-mask",
        "baseline": 50,
        "treatment": 100,
        "rationale": "Response A fully satisfies all rubric requirements, while Response B omits denoising_strength and does not show saving the output image."
      }
    ]
  }
}