EvoAgentBench: Benchmarking Agent Self-Evolution via Ability Transfer
2026-07-06 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors created EvoAgentBench to better test how AI agents improve by reusing learned procedures, not just facts. This benchmark looks at how agents transfer useful skills across different tasks in areas like web research and software development. Unlike previous tests, EvoAgentBench measures if agents can apply their past problem-solving steps to new challenges. They found that while some skill transfer happens, current methods don’t consistently improve performance. This work helps shift focus toward understanding how AI agents learn and use experience over time.
agent self-evolutionprocedural reusetransfer learningagent benchmarksAbility Graphstask execution traceslong-horizon tasksmodel backbonesexperience encodingsoftware engineering
Authors
Xingze Gao, Chuanrui Hu, Hongda Chen, Pengfei Yao, Zhao Wang, Yi Bai, Zhengwei Wu, Yunyun Han, Xiaofeng Cong, Jie Gui, Yafeng Deng, Teng Li
Abstract
Agent self-evolution in long-horizon LLM systems is largely procedural: useful experience is not merely stored information, but reusable procedures for searching, debugging, and verification. Yet current evaluations do not isolate this form of transfer. Agent benchmarks test single-episode task solving; memory benchmarks target information retention rather than procedural reuse. We introduce EvoAgentBench, a benchmark for agent self-evolution via Ability-guided transfer across four agentic domains: web research, algorithmic reasoning, software engineering, and knowledge work. EvoAgentBench extracts trace-grounded Abilities from agent executions, canonicalizes them into operational units, and builds domain-specific Ability Graphs linking tasks that share procedural overlap. By design, every test task is backed by verified training-side Ability support. Across a 528/267 train/test split, two scaffolds, and three backbones, curated Ability content transfers reliably across model families, but no current automatic method sustains positive gain in all settings. EvoAgentBench shifts self-evolution evaluation from aggregate accuracy comparison to fine-grained diagnosis of experience encoding, routing, and uptake. The benchmark is publicly available at https://huggingface.co/datasets/EverMind-AI/EvoAgentBench.