/claim #13

Summary

Adds citation-context-fit-assistant, a self-contained AI-Assisted Research Tools MVP slice that validates citation recommendations before one-click insertion into a manuscript.

The assistant evaluates whether a candidate reference:

• directly supports the highlighted claim
• contradicts a claim that was requesting support
• only contextualizes/backgrounds the claim
• is irrelevant to the highlighted claim
• has sufficient evidence strength and field fit
• needs a contrast note or intent relabeling before insertion
• is stale for a non-background citation use

Non-overlap

This is not a broad AI tools suite, evidence-grounded summarizer, citation provenance module, citation metadata integrity assistant, citation style normalizer, citation diversity auditor, citation retraction watch, methods reproducibility assistant, figure/table evidence auditor, protocol deviation screener, novelty overlap auditor, manuscript similarity triage, ethics/data availability auditor, statistical consistency checker, study power checker, unit consistency assistant, supplementary readiness assistant, or biomethods specimen/reagent provenance assistant. It focuses specifically on citation context fit and misleading one-click insertion prevention.

Local validation

Run from citation-context-fit-assistant/:

npm run check
npm test
npm run demo
npm run demo:video

All four commands passed locally.

Reviewer artifacts

• reports/summary.json
• reports/reviewer-packet.md
• reports/summary.svg
• reports/demo.webm

Safety

All data is synthetic. The module does not call DOI registries, Crossref, PubMed, arXiv, Semantic Scholar, publishers, external corpora, or live citation insertion systems. It does not include private manuscripts, credentials, real literature metadata, or live citation mutations.