celsus
github.com/0ywfe/celsusLLM workflow for automated replication checking: reads an arXiv paper's LaTeX, extracts every reported number into a typed spec, rebuilds the analysis from public data in a sandboxed run, and issues a plain-language verdict (holds up / partly / does not / could not check); model-agnostic via any chat-completions endpoint, defaults to Meta Muse (~35 min, ~$1.30/run; 7+ stars, AGPL-3.0, 2026)
Sourced from
- Awesome AI for Science — github.com/0ywfe/celsus
Related resources
Automated code generation from machine learning research papers into runnable implementations (4.5K+ stars, 2025)
Multi-agent system automatically transforming research papers into interactive AI agents with MCP server generation, tutorial auto-detection, and benchmark extraction (2.2K+ stars, MIT License, 2025)