Reproduce “ToolSched: Resource-Efficient Tool Scheduling for Shared-Host LLM Agents” 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

Shucheng Zou, Yuhao Gu, Long Chen, Zhongchun Zheng, BinZe Xie, Xianwei Zhang

ToolSched schedules shell commands from multiple agents on a shared host, queuing and staggering starts according to memory, CPU, and I/O demands. Its MCP server samples resource use, corrects declared demand estimates, and can retry commands after resource errors. On one host and 75 selected Terminal Bench 2.0 tasks, it completed 75 tasks versus 64 for unscheduled execution, in similar total time.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.