What problem does it solve? Academic papers often contain fabricated citations, fake DOIs, mismatched experiment model lists, and broken LaTeX references that undermine credibility. This Skill automates the detection and remediation of these quality defects before submission. ## Core Features & Use Cases - Fake Citation Remediation: Detects fabricated references and fake DOIs, replaces them with verified real DOIs via CrossRef, and produces a traceable replacement log mapping each original DOI to its replacement. - Experiment Claim Alignment: Cross-checks the model list claimed in the paper against benchmark suites (e.g., helix_benchmark vs notebook estimators) and itemizes deviations such as missing LightGBM or CatBoost runs. - P0 Defect Remediation Loop: Handles critical defects like fabricated numbers through a full closed loop of pattern selection, batch citation repair, and dataset unification, with sed-based fallback when LaTeX citation replacement fails. - Use Case: Given a paper directory containing .tex, PDF, and references with 3 fake DOIs and missing benchmark models, the Skill validates inputs, replaces all fake DOIs with a traceable log, aligns the experiment list, and finishes with a clean pdflatex build (0 errors, 0 undefined references) plus four standard audit reports. ## Quick Start Audit the paper in my project directory, replace any fake DOIs with verified ones, align the experiment model list with the benchmark, and produce the standard audit reports.
