What problem does it solve? AI-assisted academic writing frequently produces fabricated or inaccurate citations (roughly 40% error rate), which can lead to paper rejection, retraction, and reputational damage. This Skill provides verification principles and workflows to confirm every citation actually exists and matches its claimed metadata before it enters a bibliography. ## Core Features & Use Cases - Proactive Verification Workflow: Search each paper via WebSearch, confirm existence on Google Scholar, and copy BibTeX directly from the source instead of generating citations from memory. - Four-Layer Verification Reference: Documents format validation, existence checks, information matching (title, authors, year), and claim verification with configurable similarity thresholds. - API Reference Implementations: Includes Python clients for CrossRef, arXiv, and Semantic Scholar with rate limiting, retries, and a BibTeX/LaTeX format checker for batch or automated use cases. - Use Case: While writing an ML paper with the ml-paper-writing skill, every new citation is automatically verified on Google Scholar; unverifiable entries are marked as [CITATION NEEDED] and reported to the user. ## Quick Start Ask the assistant to verify the citations in your paper's BibTeX file using Google Scholar and flag any references that cannot be confirmed.
