Grandma's Wisdom — Academic Claim & Citation Auditor ⭐

The problem

Large language models often hallucinate academic citations or overstate the findings of legitimate research. Relying on AI-generated claims without verification leads to the spread of misinformation and undermines professional credibility.

What it does

• Decomposes complex academic statements into atomic, testable claims.
• Cross-references citations against DOIs, journal titles, and author metadata to identify fabrications.
• Evaluates claim-to-citation alignment to detect overclaiming, causal leaps, and population generalization errors.
• Scores evidence strength using a tiered framework from meta-analyses to SEO content farms.
• Rewrites unsupported or misleading claims into evidence-calibrated language suitable for publication.

Why this beats prompting it yourself

General prompts often suffer from authority theater, where the agent accepts a prestigious institution name as proof of truth. This skill enforces a rigorous audit trail that treats citations as metadata until the specific context, methodology, and effect sizes are verified.

Use cases

• Auditing AI-generated research papers or technical whitepapers for factual accuracy.
• Verifying the legitimacy of citations in marketing copy that references scientific studies.
• Refactoring overstated business claims into legally defensible, evidence-backed statements.
• Detecting "hallucinated" DOIs and non-existent papers in academic drafts.

Known limitations

Cannot verify claims if the cited source is behind a paywall and not provided in the context. Distinguishes between unavailable evidence and contradictory evidence without assuming falsehood.