Source record

Published by arXiv cs.MA (T1 source). The original is at https://arxiv.org/abs/2610.00980.

Pipeline notes

The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.

The paper introduces Runtime Agent Coordination (RAC), which selects agents from existing AI-scientist hosts during execution, assigns scoped work contracts, and provides artifact-grounded verification. Evaluated on ResearchClawBench across Agent Laboratory, EvoScientist, and ARK, runtime selection improved mean scores, but adding contracts and verification reduced them under host-calibrated budgets.

Tests runtime vs design-time orchestration for multi-agent AI scientists with open code. Mixed results — selection helps, contracts and verification hurt — give a concrete data point against premature coordination complexity.

Cited by

No citations on record.