This is a complete autonomous research system that takes a topic string and outputs a conference-ready LaTeX paper with real citations from arXiv and Semantic Scholar. It runs a 23-stage pipeline including literature review, multi-agent hypothesis formation, sandboxed Python experiments with auto-healing, peer review simulation, and a four-layer citation verification pass to prevent hallucinations. You can run it fully hands-off with auto-approve or pause at three gate stages to review literature, experimental design, and final quality. The output directory includes the paper in both Markdown and LaTeX, a pruned BibTeX file, experimental code and results, and a knowledge base for cross-run learning. Works with OpenAI, OpenRouter, or Agent Client Protocol.
npx skills add https://github.com/aradotso/trending-skills --skill autoresearchclaw-autonomous-research