/claim #16

Summary

• Adds bayesian-prior-sensitivity-assistant, a dependency-free synthetic auto peer-review module for Bayesian prior sensitivity and posterior robustness.
• Flags undisclosed priors, unjustified informative priors, posterior direction flips under alternate priors, large posterior shifts, convergence failures, divergent transitions, low effective sample size, weak prior predictive checks, and strong claims that are not robust across priors.
• Generates reviewer artifacts under bayesian-prior-sensitivity-assistant/reports/, including demo.avi.

Validation

• npm run check
• npm test
• npm run demo
• npm run demo:video
• git diff --check
• git diff --cached --check

Demo

The short demo video is committed at bayesian-prior-sensitivity-assistant/reports/demo.avi.

Scope Boundary

This is intentionally separate from broad AI assistant suite work, multiple-comparison control, generic statistical consistency checks, study power feasibility, uncertainty tone review, prompt safety, literature freshness, image integrity, external validity, and protocol trace workflows. It focuses only on Bayesian prior sensitivity and posterior robustness as an auto peer-review red-flag layer.