docs(score-tests): canonical inline rubric aligned with scoreFile() - #7
Conversation
…) (F1/F2) Paraphrase deleted quality-scorecard into the score-tests.js comment block (penalty tables, thresholds, limitations). Fix README fixture penalty claim to match implemented −5 for direct @playwright/test imports. Verification: node --check tools/score-tests.js; F2 falsifier grep clean. Co-authored-by: Cursor <cursoragent@cursor.com>
SDK code review (trainer → form-check + review-rigor)Verdict:
Summary
Findings (ranked)F1 · P4 · COR · conf 72%Residual internal review marker in production source Round-1 remediation left a permanent // --- Maintainability (-20 max) ---
// sdk-review F1: fixture-import check lives here to match rubric comment taxonomy
const isSpecFile = /\.(spec|test)\.(ts|js)$/.test(filePath);Downstream consequence: future contributors/agents may treat SDK review IDs as durable architecture docs. Remove in a follow-up cleanup commit; not merge-blocking because behavior and rubric alignment are correct. Scorecard (S1–S7): S1 repro ✓ · S2 cite ✓ · S3 scope ✓ · S4 severity proportionate ✓ · S5 fix path clear ✓ · S6 no false COR on behavior ✓ · S7 merge impact none ✓ → 7/7 F2 · P4 · ARC · DRIFT · conf 68%CLI banner still says "Quality Scorecard" after scorecard markdown purge Human-readable output header at Scorecard: 6/7 (S6: pre-existing, out of diff scope — still note for agent confusion) F3 · P4 · COR · conf 65%Fixture-import detector matches literal
Scorecard: 6/7 Merge recommendationMerge. Round-1 P2/P3 items (README fixture lie, Maintainability taxonomy drift, CHANGELOG internal IDs) are resolved. Ship threshold for docs-only (≥85%): met with zero P1–P3 blockers. Optional follow-ups: strip Posted by |
…ures path - F1: fixture-import check under Maintainability; drop internal IDs from CHANGELOG - F2: CLI output header "Quality Rubric" (was Quality Scorecard after references purge) - F3: exempt ../fixtures imports alongside ./fixtures (GETTING_STARTED layout) Co-authored-by: Cursor <cursoragent@cursor.com>
f264336 to
579d4c3
Compare
SDK codereview — round 2 (F2·P4, F3·P4)Verdict: APPROVE — cosmetic follow-ups from
F1·P4 (residual Verify: |
Summary
Automated SDK weekend worker +
_sdk_verify_and_pr.shverification.SDK code review (trainer → form-check + review-rigor)
Verdict:
APPROVEQueue:
playwrighter-rubric-alignBranch:
chore/purge-references-sdk· Repo:weijia-89/playwrighterlocalonly/sdk-reviews/playwrighter-rubric-align-chore-purge-references-sdk.mdcursor-sdk-playground/prompts/reviews/playwrighter-rubric-align-chore-purge-references-sdk.mdSummary
tools/score-tests.js:139–152); Unreleased CHANGELOG no longer embeds internal review IDs (CHANGELOG.md:15).scoreFile(): direct@playwright/testimports in spec files cost 5 points;test.step()/ per-test length remain non-scored guidance (README.md:30).references/scorecard; comment taxonomy aligns withscoreFile()implementation (tools/score-tests.js:34–77,89–226).Full ranked findings are in the PR review comment posted by the SDK hook.
Posted by
_sdk_verify_and_pr.shafter_sdk_trainer_codereview.sh. Merge still requires CI green and human review;BLOCK/REQUEST_CHANGES(withSDK_CODEREVIEW_STRICT=1) stop the hook before push.Test plan
test ! -d references && ! rg -n 'references/quality-scorecard|references/RESEARCH_INDEX|references/ADVERSARIAL' --glob '!localonly/**' --glob '!CHANGELOG.md' . && node --check tools/score-tests.js && ! rg -n 'uses fixtures|describe/group|test\.step or single' tools/score-tests.jsSDK_CODEREVIEW_SKIP=1)