{
  "claim": "The public CodeSkillBank release is internally consistent on its totals, and several of its details differ from the paper. Every count below comes from reading the released parquet, the release's own metadata, or the Hugging Face and SWE-bench listings. None of it comes from the paper's prose.",
  "method": "I downloaded all 19 edges shards and read the 16 cards shards one at a time, extracting only provenance and text fields. edges.repo_name holds a GitHub owner; the repository is the first segment of relative_path in each card's examples, and examples are capped at 8 per card. The pattern_annotations shards were read for model and label_source. Language counts go by file extension. 'Vendored' means a path under node_modules/, vendor/, third_party/, external/, deps/ or similar, or a file named like jquery.js or bootstrap.js. It is a heuristic, not an audit. SWE-bench Verified's repo counts come from the Hugging Face datasets-server statistics for princeton-nlp/SWE-bench_Verified. I executed no third-party code.",
  "source": "https://huggingface.co/datasets/ant-intl/DeveloperSkills-Code2Skill",
  "captured": "2026-09-26",
  "note": "Rows 1 and 2 confirm the paper. Rows 3 to 6 are the scale arithmetic. The rest are things the paper does not say.",
  "columns": [
    { "key": "check", "label": "check" },
    { "key": "value", "label": "result", "align": "right", "mono": true },
    { "key": "reading", "label": "what it means" }
  ],
  "rows": [
    { "check": "edges rows + dropped rows (metadata)", "value": "945,993 + 60,829 = 1,006,822", "reading": "equals the paper's accepted count exactly" },
    { "check": "cards by granularity", "value": "490,457 / 233,549 / 26,530 / 212", "reading": "atomic, composite, mixed, pattern-only; the paper's Figure 8 omits the 212 and totals 750,536, not 750,748" },
    { "check": "distinct owners in edges", "value": "6,103", "reading": "the unreferenced source figure says 6,115 'rep groups'" },
    { "check": "distinct owner/repo in card examples", "value": "7,399", "reading": "a lower bound; the examples cover 918,083 of 945,993 records (97.0%)" },
    { "check": "records per owner: median / mean / max", "value": "124 / 155.0 / 448", "reading": "alibaba: 31 repos, 246 records; facebook: 13 repos, 65 records" },
    { "check": "records by language family", "value": "8 families", "reading": "Python 236,827; C++ 121,559; TypeScript 114,471; C 100,293; JavaScript 95,438; Go 90,715; Java 86,603; Rust 72,177; no Kotlin, C#, Swift, PHP or Ruby" },
    { "check": "cards whose text contains trivial, getter or setter", "value": "0 of 750,748", "reading": "the low-value regex ran; it also matches 'non-trivial'" },
    { "check": "task_family = validation", "value": "348,612 (46.4%)", "reading": "the paper's 'validation (46.4%)'; testing appears on 10,762 cards, a family in neither the paper's vocabulary nor the reference code" },
    { "check": "largest purpose card", "value": "761 records, 398 owners", "reading": "readArguments: command-line argument parsing" },
    { "check": "example paths that look vendored", "value": "16,701 (1.8%)", "reading": "for example Bootstrap's Toast._maybeScheduleHide, credited to BuffaloWill/oxml_xxe" },
    { "check": "cards reviewed for patterns by a model", "value": "14,539", "reading": "all by DeepSeek-V4-Flash-0731; 3,600 confirmed; 736,209 settled by deterministic gates" },
    { "check": "SWE-bench Verified source repos present", "value": "2 of 12", "reading": "django/django 109 records, astropy/astropy 260; together 253 of the 500 Verified instances" },
    { "check": "AdguardTeam/dnsproxy (the paper's Figure 2) in examples", "value": "0", "reading": "AdGuard's 280 records list AdGuardHome and AdguardForiOS only" },
    { "check": "dropped parquet files", "value": "0 of 2", "reading": "RELEASE_STATUS.json reports 2 files and 60,829 rows; the repository has none" },
    { "check": "datasets-server on the top-level card", "value": "no data files", "reading": "configs point at data/cards/*.parquet; the files live under codeskillbank_hf_release_with_patterns_20260814/" },
    { "check": "licence on the dataset card", "value": "apache-2.0", "reading": "the inner README and LICENSE_REVIEW.md both say license: other until a provenance audit is done" }
  ]
}