数学建模竞赛独立盲审 Codex Skill:需求覆盖、模型与数值审计、复现、评分和合规检查
-
Updated
Sep 11, 2026 - Python
数学建模竞赛独立盲审 Codex Skill:需求覆盖、模型与数值审计、复现、评分和合规检查
Evidence-bound thesis revision, professor-style review, corpus strategy mining, and Word output skill.
Local Windows AI prompt enhancer (Tauri 2 + Rust). 100% offline: prompts, history, API key stay on disk (SQLCipher). Quantitative effect numbers are not published (v0.3.2 evaluation: 60 samples x 4 targets x 3 repeats + 48-unit blind human review found no quality advantage for enhancement; automated eval biases toward longer/structured outputs).
Author-persona tournament: a reproducible, topic-luck-corrected standard for blind paper review (Ralphthon @ ICML 2026)
An interactive paper anonymizer for PDF and DOC/DOCX files with drag-to-select redaction, metadata removal, and precise word-level editing.
Evidence-backed, blind AI reviewer that scores candidate code repositories. Advisory and human-in-the-loop: runs profile-specific checks in sandboxed Docker workers, cites source-level evidence, and supports resume/GitHub enrichment plus batch CSV evaluation.
PRISM workflow and systematic literature web
A blind, fair, local DSH Web arena: same task, isolated worktrees, shared verification, judge before reveal.
To associate your repository with the blind-review topic, visit your repo's landing page and select "manage topics."