Task 039: Expand Cross-Episode Analysis Beyond Episode-Based Structure

Status: DONE

Motivation

Human directive: "クロスエピソード分析はそれ単体だけでなく、各船の仕様など、エピソード以外の観点からも柔軟に行ってよい"

The current cross-episode analysis (reports/data/summary/cross-episode.json) is structured around comparing metrics across episodes. The human wants additional analysis dimensions:

• Ship specifications (Kestrel mass/thrust evolution, damage accumulation)
• Physical science accuracy (real vs. depicted values)

Implementation (Codex-reviewed design)

New Types (report-types.ts)

• TimelineEvent: episode, timestamp, label, description, stateChanges, evidenceQuoteId
• EventTimeline: caption, events[] — for ship damage/repair history
• VerificationStatus: "verified" | "approximate" | "unverified" | "discrepancy"
• VerificationRow: claim, episode, depicted, reference, source, accuracyPercent, status
• VerificationTable: caption, rows[] — for science accuracy scorecard
• SummarySection: extended with optional eventTimeline and verificationTable

New Renderers (templates.ts)

• renderEventTimeline(): vertical timeline with markers, state changes, episode badges
• renderVerificationTable(): table with Japanese headers, accuracy %, status badges
• verificationStatusInfo(): maps status to Japanese label + CSS class
• CSS: .event-timeline, .timeline-track, .timeline-event, .verification-table, .verification-badge, status colors

Ship Technical Dossier (ship-kestrel.json)

• Generator: ship-kestrel-analysis.ts + generate-ship-kestrel.ts
• Tests: ship-kestrel-analysis.test.ts (17 tests)
• Sections: 基本仕様, 損傷・修復の経過 (with EventTimeline), 推力・加速度の変遷, 質量の謎, 限界パラメータのドラマツルギー, 総合評価

Physics & Science Accuracy (science-accuracy.json)

• Generator: science-accuracy-analysis.ts + generate-science-accuracy.ts
• Tests: science-accuracy-analysis.test.ts (23 tests)
• Sections: 検証方法, 検証スコアカード (with VerificationTable), Brachistochrone力学の検証, 実測データとの照合, 航法精度の検証, 総合評価
• Verification scorecard: 11 items, 7 verified, 3 approximate, 1 unverified, 0 discrepancy

Test Results

• 814 TS tests + 79 Rust tests = 893 total (0 failures)
• Site builds: 5 episodes, 24 transfers, 3 summaries, 7 logs

Depends on

• Cross-episode analysis infrastructure (Task 021)
• All episode analyses