{"id":"arxiv:2605.13821","title":"Harnessing Agentic Evolution","date":"2026-05-13","section":"Agents, open-endedness, AGI","quality":0.05538,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/","source_url":"https://arxiv.org/abs/2605.13821","text_available":true,"authors":["Jiayi Zhang","Yongfeng Gu","Jianhao Ruan","Maojia Song"],"outgoing":[{"source":"arxiv:2605.13821","target":"arxiv:2408.08435","evidence":"id","id":"arxiv:2408.08435","title":"Automated Design of Agentic Systems","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c9890d5668d1c1b33e29569b74626fd586f456af6a97c657b8c99895c90f59a5/"},{"source":"arxiv:2605.13821","target":"arxiv:2505.22954","evidence":"id","id":"arxiv:2505.22954","title":"Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c29f716871f80314868b5647692de387f26e12166b93bd362915fd272b6e081c/"},{"source":"arxiv:2605.13821","target":"arxiv:2507.19457","evidence":"id","id":"arxiv:2507.19457","title":"GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c1ef72d96fa3292590a6a3d112ab509840081b59aaa771cf42b8b0e8ef62ba08/"},{"source":"arxiv:2605.13821","target":"arxiv:2512.18746","evidence":"id","id":"arxiv:2512.18746","title":"MemEvolve: Meta-Evolution of Agent Memory Systems","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/c2161237fe9d9768afaa2880a40019455779801af5fba05be5781c229c18ea86/"},{"source":"arxiv:2605.13821","target":"arxiv:2601.16175","evidence":"id","id":"arxiv:2601.16175","title":"Learning to Discover at Test Time","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/5d0b45cbbc72945765e689cbfb7ac7e4e92dfbb764861b27682ac0744ef05b70/"},{"source":"arxiv:2605.13821","target":"arxiv:2602.07755","evidence":"id","id":"arxiv:2602.07755","title":"Learning to Continually Learn via Meta-learning Agentic Memory Designs","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/4d0a0ea160495ea50583e0e24027b6085e2f430a44195e3053ba69cda8cc82ea/"},{"source":"arxiv:2605.13821","target":"arxiv:2603.19461","evidence":"id","id":"arxiv:2603.19461","title":"Hyperagents","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/17eec8a15debc212b5530f117dd8ba8bad339590729cbbc4f75a47755ebcb544/"},{"source":"arxiv:2605.13821","target":"arxiv:2604.19341","evidence":"id","id":"arxiv:2604.19341","title":"Structured Scaling of AI Discovery Across Diverse Scientific Domains","url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/33ad3375247f86197210840a19064548b997e8617475e606d7a7ac2cda64740e/"}],"incoming":[],"text_parts":[{"start":0,"end":15937,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html"},{"start":15937,"end":31637,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html"},{"start":31637,"end":41657,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html"}],"contents":[{"title":"Harnessing Agentic Evolution","offset":0,"level":1,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"Abstract","offset":32,"level":6,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"1 Introduction","offset":1599,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"2 Related Work","offset":7002,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"3 Problem Formulation","offset":8978,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"3.1 Agentic Evolution","offset":9004,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"3.2 Evolution as an Interactive Environment","offset":11729,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"4 Methodology","offset":14481,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"4.1 Design of AEvo","offset":15118,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-1.html#text"},{"title":"4.2 Instantiating AEvo","offset":17933,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"5 Experiments","offset":19468,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"5.1 Settings","offset":19486,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"5.2 Main Results","offset":21782,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"5.3 Evolution Dynamics","offset":24676,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"5.4 Ablation Study","offset":26217,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"5.5 Case Study","offset":27139,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"6 Conclusion","offset":28511,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"Appendix A Ablation Study Details","offset":29734,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"Appendix B Implementation Details","offset":29772,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"Appendix C Additional Experimental Details","offset":30499,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"C.1 Task and Metric Details","offset":30546,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"Circle Packing.","offset":30987,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"Autocorrelation-Second.","offset":31208,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"Performance Engineering.","offset":31605,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-2.html#text"},{"title":"Reporting Protocol.","offset":31947,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Cost Computation.","offset":32468,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"C.2 Initialization Details","offset":33144,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"C.2.1 Initial Procedure in Procedure-Based AEvo","offset":33176,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"C.2.2 Initial Agent in Agentic and Reasoning Tasks","offset":33768,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"C.2.3 Meta-Agent Skill Excerpt","offset":34307,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"C.3 Representative Evolved Artifacts and Outcomes","offset":34718,"level":3,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"C.3.1 Representative Evolved Procedures","offset":34773,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"C.3.2 Representative Evolved Agent Harness","offset":35335,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Task skill.","offset":35784,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Session goal.","offset":35806,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Persistent family map.","offset":35830,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Replay utility.","offset":35863,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Session memory written back by the inner agent.","offset":35889,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"C.3.3 Representative Optimization Outcomes","offset":36274,"level":4,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"ARC-AGI-2 best artifact.","offset":36323,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Analysis.","offset":36737,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Performance-engineering best artifact.","offset":37841,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Submitted wrapper.","offset":38337,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Parameterized benchmark family.","offset":38366,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Core benchmark schedule.","offset":38408,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Analysis.","offset":38449,"level":5,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"},{"title":"Instructions for reporting errors","offset":40693,"level":2,"url":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/part-3.html#text"}],"text_code_points":41657,"source_markdown":"https://vinnik-dmitry07.github.io/favorite-papers/agents/papers/1f1060a507e2bf439d2934d610827d1b3177d85c90c1a2cab483608a56de5138/source.md"}