/claim #16

Summary

• Adds conflict-disclosure-consistency-assistant, a self-contained research-assistant slice for conflict-of-interest and funding-disclosure consistency before peer-review packets are shown.
• Compares author affiliations, funders, grants, sponsor roles, trial-registry funding, restricted data-provider relationships, and related-party primary-support citations against the reviewer-visible disclosure packet.
• Generates deterministic JSON, Markdown, SVG, and H.264 MP4 demo artifacts from synthetic project metadata.

Why this solves part of #16

• The issue asks for auto peer-review reports, methodological red flags, missing citations or scope misalignment, claims-vs-evidence alignment, and reproducibility confidence. Hidden sponsor, funder, author, data-provider, or citation relationships directly affect those review outcomes.
• This module blocks release of an assistant review packet when disclosures are incomplete or inconsistent, then emits concrete revision tasks that authors can fix before submission.

Acceptance and reviewer map

• Auto peer review reports: reviewer actions and disclosure-focused findings are emitted in JSON and Markdown.
• Methodological red flags: sponsor role, funder overlap, and author relationship gaps are escalated as critical review blockers.
• Missing citations or scope misalignment: related-party primary-support citations are flagged when context is absent.
• Claims-vs-evidence alignment: funding and data-provider relationships are attached to evidence review before the packet is released.
• Reproducibility context: trial registry funding and restricted data-provider terms are checked against disclosure metadata.

Sponsor decision packet

• What the bounty poster gets: a complete disclosure-consistency review path with hold/revise/ready decisions, blocker codes, author impact, undisclosed funding math, reviewer actions, and stable audit digest output.
• How to verify it: rerun the listed Node checks and inspect reports/summary.json, reports/disclosure-review-packet.md, reports/summary.svg, and the committed demo video. The expected synthetic case is hold_disclosure_review with 3 reviewed projects, 1 held project, 2 ready projects, 1 affected author, 5 critical findings, 3 warnings, and $250,000 in undisclosed funding.
• Why this submission should be rewarded: existing #16 submissions cover broad assistant flows, evidence/protocol traces, statistics, research-gap planning, rebuttal packs, ethics/data availability, citation-context, reporting guidelines, benchmark leakage, figure/table consistency, analysis-variable provenance, domain templates, grant-fit, limitations, uncertainty, supplement readiness, prompt safety, and study power. This submission covers the distinct disclosure-integrity layer that determines whether the assistant packet itself is fair and reviewer-safe.

Non-overlap

• This is separate from broad assistant suites, protocol/evidence traces, statistics review, research-gap planning, rebuttal response packs, ethics/data availability checks, citation-context reconciliation, reporting-guideline checks, benchmark leakage, figure/table checks, analysis-variable provenance, domain-template selection, grant-fit scoring, limitations disclosure, uncertainty calibration, supplement readiness, prompt safety, and study-power feasibility.
• Scope is limited to conflict-of-interest, funding, sponsor-role, data-provider, trial-registry, and related-party citation disclosure consistency.

Scope boundaries

• Does not score the scientific validity of the manuscript, execute notebooks, call external services, choose reviewers, or replace a full ethics review.
• Intentionally focuses on whether the assistant-generated peer-review packet can be released with adequate disclosure context.

Risk and regression notes

• The module is dependency-free and synthetic-data-only, so it introduces no service, credential, payment, privacy, or database risk.
• The guard is conservative by design: missing sponsor role, funding mismatch, author relationship gaps, or incomplete disclosure sections produce hold/revise tasks rather than a misleading ready status.
• Existing project files are untouched; the change is isolated under one new directory and can be reviewed or removed without affecting the rest of the repository.

Validation

• cd conflict-disclosure-consistency-assistant && npm run check
• cd conflict-disclosure-consistency-assistant && npm test
• cd conflict-disclosure-consistency-assistant && npm run demo
• cd conflict-disclosure-consistency-assistant && file reports/demo.mp4 reports/summary.svg reports/summary.json reports/disclosure-review-packet.md
• cd conflict-disclosure-consistency-assistant && mdls -name kMDItemCodecs -name kMDItemDurationSeconds -name kMDItemPixelHeight -name kMDItemPixelWidth reports/demo.mp4
• git diff --check

Demo and artifacts

• Short demo video: conflict-disclosure-consistency-assistant/reports/demo.mp4
• Review summary: conflict-disclosure-consistency-assistant/reports/summary.json
• Reviewer packet: conflict-disclosure-consistency-assistant/reports/disclosure-review-packet.md
• Visual summary: conflict-disclosure-consistency-assistant/reports/summary.svg