Giving an enterprise AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → a command shell can outperform supplying a catalog of specialized functions. A study using Anthropic's Opus 4.8 and OpenAI's GPT-5.5 found higher task scores with Bash alone on two workplace benchmarks, while using fewer model tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition →. A shell lets the agent combine commands and process data within one execution. The paper compares these interfaces while holding the underlying model and task constant.
Adding specialized tools to Bash produced no consistent quality gain. The authors favor shell access where execution can be isolated, and programs restricted to approved functions where organizational policy requires tighter controls.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
tool use — A model's ability to call external functions — run code, search the web, edit files — instead of only generating text.
articleThe Devil Is in the Interface: Evaluating How Tool Architecture Shapes Coding Agent BehaviorXiangzhe Xu, Hamidreza Saghir, Qianhui Wu, Marc-Alexandre C\^ot\'e, Tong Wang, Kiran Lakkaraju, Kexin Pei, Xiangyu Zhang
articleHPCAgent-Bench: A Framework for Sound Benchmarking of Code-Optimizing Agents, Harnesses, and SkillsYakup Koray Budanaz, Lorenzo Paleari, Marcin Copik, Alexandru Calotoiu, Timo Schneider, Benjamin Chen Ming Choong, Shriram Chandran, Lars Hulsbergen, Levi Widmer, Torsten Hoefler
articleHarness Engineering in LLM Tool Use via Agent-Native Reusable Tool PrimitivesHaibo Jin, Suijin Wang, Xucheng Yu, Haojing Luo, Haohan Wang