AI & ML New Capability

Introduces a decentralized, multi-agent framework for scientific discovery that uses an 'ArtifactReactor' for plannerless coordination and full computational lineage.

arXiv · March 17, 2026 · 2603.14312

Fiona Y. Wang, Lee Marom, Subhadeep Pal, Rachel K. Luu, Wei Lu, Jaime A. Berkovich, Markus J. Buehler

The Takeaway

This moves beyond simple LLM-based researchers to a system where independent agents autonomously discover and fulfill each other's data needs via a shared DAG (Directed Acyclic Graph). It provides a structured, auditable architecture for scaling agentic scientific investigation without central coordination.

From the abstract

We present ScienceClaw + Infinite, a framework for autonomous scientific investigation in which independent agents conduct research without central coordination, and any contributor can deploy new agents into a shared ecosystem. The system is built around three components: an extensible registry of over 300 interoperable scientific skills, an artifact layer that preserves full computational lineage as a directed acyclic graph (DAG), and a structured platform for agent-based scientific discourse