HyperTool: Beyond Step-Wise Tool Calls for Tool-Augmented Agents
For LLM agents performing multi-step tool use, HyperTool addresses the execution-granularity mismatch that wastes context and forces models to manage low-level dataflow.
HyperTool introduces a unified executable interface that replaces step-wise atomic tool calls with a single code block, improving multi-step tool use accuracy from 15.69% to 35.29% on Qwen3-32B and from 9.93% to 33.33% on Qwen3-8B on the MCP-Universe benchmark.
Tool-augmented LLM agents commonly rely on step-wise atomic tool calls, where each invocation, observation, and value transfer is exposed in the main reasoning trace. This creates an \emph{execution-granularity mismatch}: locally deterministic tool workflows are unfolded into repeated model-visible decisions, consuming context and forcing the model to manage low-level dataflow in the trace. We introduce \textbf{HyperTool}, a unified executable MCP-style tool interface that changes the model-visible unit of tool execution. A model invokes HyperTool with a code block that can call existing tools through their original schemas, manipulate returned values, and pass intermediate results locally, folding deterministic tool subroutines into a single outer call. To train models to use this interface, we synthesize HyperTool-format trajectories from cross-tool compositional tasks and verify them in real MCP environments. On MCP-Universe, HyperTool improves average accuracy from 15.69\% to 35.29\% on Qwen3-32B and from 9.93\% to 33.33\% on Qwen3-8B, and surpass GPT-OSS and Kimi-k2.5 on average accuracy, showing that our HyperTool can substantially improve multi-step tool use.