EvoTool: Self-Evolving Tool-Use Policy Optimization in LLM Agents via Blame-Aware Mutation and Diversity-Aware Selection

TL;DR

EvoTool introduces a modular, evolutionary approach to optimize tool-use policies in LLM agents, effectively addressing credit assignment issues and improving performance across multiple benchmarks.

Contribution

It presents a novel self-evolving framework with blame attribution, targeted mutation, and diversity preservation for modular policy optimization in LLM agents.

Findings

Outperforms strong baselines by over 5 points on GPT-4.1 and Qwen3-8B.

Achieves superior efficiency and transferability.

Effectively localizes failures and improves modules through natural-language critiques.

Abstract

LLM-based agents depend on effective tool-use policies to solve complex tasks, yet optimizing these policies remains challenging due to delayed supervision and the difficulty of credit assignment in long-horizon trajectories. Existing optimization approaches tend to be either monolithic, which are prone to entangling behaviors, or single-aspect, which ignore cross-module error propagation. To address these limitations, we propose EvoTool, a self-evolving framework that optimizes a modular tool-use policy via a gradient-free evolutionary paradigm. EvoTool decomposes agent's tool-use policy into four modules, including Planner, Selector, Caller, and Synthesizer, and iteratively improves them in a self-improving loop through three novel mechanisms. Trajectory-Grounded Blame Attribution uses diagnostic traces to localize failures to a specific module. Feedback-Guided Targeted Mutation then edits only that module via natural-language critique. Diversity-Aware Population Selection preserves complementary candidates to ensure solution diversity. Across four benchmarks, EvoTool outperforms strong baselines by over 5 points on both GPT-4.1 and Qwen3-8B, while achieving superior efficiency and transferability. The code will be released once paper is accepted.