Reproduce “Decentralized Multi-Agent Coordination and Communication in Shared Codebases” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Rishabh Mediratta, Lingyun Xiao, Peter Stone, Joydeep Biswas, Elias Stengel-Eskin
ParallACT evaluates whether independently assigned coding agents can integrate edits in shared codebases. Its two-agent benchmark compares parallel agents seeing only their own requests with an oracle given both. On 50 contradictory instances, joint pass rate is 62% for the oracle and 16% for parallel agents. These results identify a coordination gap when such challenges are present, not their prevalence generally.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.