/claim #16

Summary

• Add manuscript-claim-evidence-alignment/, a distinct AI research assistant slice for checking manuscript claims against cited evidence before release.
• Classify claims as supported, weakly supported, overclaimed, or missing evidence using required evidence signals such as p-values, sample size, control groups, safety monitoring windows, and novelty scope limitations.
• Emit deterministic reviewer artifacts in JSON, Markdown, SVG, and MP4 form using synthetic data only.

Scope Boundaries

This is synthetic demo data only. It does not call external AI models, manuscript services, publisher APIs, notebooks, private documents, credentials, or live SCIBASE accounts.

This scope is intentionally separate from existing #16 slices such as broad peer-review reports, reproducibility execution, research gap discovery, citation recommendation, general writing assistants, or full manuscript analysis dashboards.

Demo

• manuscript-claim-evidence-alignment/reports/demo.mp4

Validation

• node manuscript-claim-evidence-alignment/test.js
• node manuscript-claim-evidence-alignment/demo.js
• node manuscript-claim-evidence-alignment/render-video.js
• ffprobe confirmed H.264, 960x540, 5 seconds, 120 frames for the demo MP4.
• node --check manuscript-claim-evidence-alignment/index.js
• node --check manuscript-claim-evidence-alignment/sample-data.js
• node --check manuscript-claim-evidence-alignment/test.js
• node --check manuscript-claim-evidence-alignment/demo.js
• node --check manuscript-claim-evidence-alignment/render-video.js
• git diff --cached --check

No payout details, secrets, tokens, cookies, or private account information are included in this PR.