Once I wrote in the MCP tool description when to use the tool, AI agents called it
Summary I built an MCP tool that assigns multiple independent tasks to subagents and has the subagents execute them in parallel. Below, I call this MCP tool the parallel-execution MCP tool. MCP stands for Model Context Protocol, a standard for AI agents to find and call external tools. A subagent is a separate AI agent that an AI agent launches in order to delegate a task to it. However, the…
In a study examining the conditions under which AI agents call a parallel-execution Model Context Protocol (MCP) tool, researchers discovered that adding a specific sentence to the tool's description prompted the AI agents to call the tool seven out of twelve times. The sentence provided a clear instruction on when to utilize the tool.
When the sentence was absent, the AI agents did not call the tool at all. The researchers tested this hypothesis by conducting twelve separate runs, with six runs including a rule file and six runs without it. The findings suggest that the weight of the processing involved in calling the tool influences the decision to invoke it.
This phenomenon is particularly relevant for AI coding agents such as OpenHands, OpenCode, and Qwen Code.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
