Human-in-the-Loop

AutoResearchClaw is an open-source 23-stage autonomous research pipeline from UNC Chapel Hill that turns a research idea into a conference-ready LaTeX paper. Features: multi-agent debate for hypothesis generation, self-healing executor with Pivot/Refine decision loop, verifiable result reporting preventing hallucinations, human-in-the-loop with 7 intervention modes, and cross-run evolution. Outperforms AI Scientist v2 by 54.7% on ARC-Bench. MIT licensed.

Automating scientific discovery requires more than generating papers from ideas. Real research is iterative: hypotheses are challenged from multiple perspectives, experiments fail and inform the next attempt, and lessons accumulate across cycles. We present AutoResearchClaw, a multi-agent autonomous research pipeline built on five mechanisms: structured multi-agent debate for hypothesis generation and result analysis, a self-healing executor with a Pivot/Refine decision loop that transforms failures into information, verifiable result reporting that prevents fabricated numbers and hallucinated citations, human-in-the-loop collaboration with seven intervention modes, and cross-run evolution that converts past mistakes into future safeguards. On ARC-Bench, AutoResearchClaw outperforms AI Scientist v2 by 54.7%.