skills/web-scraper/_scores.json

{
  "version": "1.0.0",
  "skillHash": "sha256:c58572e737efd01c5086adce2904f86777a9982dff6d7057333173b25d79e09c",
  "scoredAt": "2026-05-13T10:35:56.220Z",
  "backend": "subagent",
  "model": "claude-sonnet-4-6",
  "quality": {
    "score": 87,
    "dimensions": {
      "clarity": "PASS",
      "completeness": "WEAK",
      "conciseness": "PASS",
      "actionability": "PASS",
      "crossPlatform": "WEAK",
      "examples": "PASS"
    },
    "issues": [
      {
        "severity": "MEDIUM",
        "category": "completeness",
        "detail": "The skill does not cover all error scenarios such as network failures, retries, or handling of robots.txt violations, limiting robustness."
      },
      {
        "severity": "MEDIUM",
        "category": "crossPlatform",
        "detail": "The skill is Python-specific and requires certain libraries, limiting its applicability to agents that can run Python code."
      }
    ]
  },
  "security": {
    "verdict": "SAFE",
    "issues": []
  },
  "impact": {
    "baselineAvg": 63,
    "treatmentAvg": 95,
    "multiplier": 1.51,
    "scenarios": [
      {
        "name": "Paginated catalog scrape with hash dedup and integer-cent prices",
        "baseline": 55,
        "treatment": 95,
        "rationale": "Baseline uses requests+bs4 and pagination but dedupes by name and stores prices as floats while treatment adds User-Agent/timeout, integer cents, and md5 content-hash dedup with separated extract/transform."
      },
      {
        "name": "JS-rendered listings via Playwright + Bright Data residential proxy",
        "baseline": 70,
        "treatment": 95,
        "rationale": "Both use headless Playwright with networkidle and a proxy, but treatment hands rendered HTML to BeautifulSoup and uses the exact customer-zone-{zone} auth on brd.superproxy.io:22225."
      }
    ]
  }
}