{"id":97718,"url":"https://github.com/EvoAgentX/Awesome-Self-Evolving-Agents","name":"Awesome-Self-Evolving-Agents","description":"[Survey] A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems","projects_count":217,"last_synced_at":"2026-08-17T20:00:26.539Z","repository":{"id":309488516,"uuid":"980086077","full_name":"EvoAgentX/Awesome-Self-Evolving-Agents","owner":"EvoAgentX","description":"[Survey] A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems","archived":false,"fork":false,"pushed_at":"2026-05-16T14:49:18.000Z","size":2338,"stargazers_count":2383,"open_issues_count":47,"forks_count":177,"subscribers_count":27,"default_branch":"main","last_synced_at":"2026-07-29T10:04:11.982Z","etag":null,"topics":["agent","agentic-ai","ai","llms","multi-agent-systems","natural-language-processing","self-evolving"],"latest_commit_sha":null,"homepage":"https://arxiv.org/abs/2508.07407","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/EvoAgentX.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-05-08T14:38:46.000Z","updated_at":"2026-07-29T09:20:18.000Z","dependencies_parsed_at":"2025-10-08T12:20:22.751Z","dependency_job_id":null,"html_url":"https://github.com/EvoAgentX/Awesome-Self-Evolving-Agents","commit_stats":null,"previous_names":["evoagentx/awesome-self-evolving-agents"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/EvoAgentX/Awesome-Self-Evolving-Agents","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EvoAgentX%2FAwesome-Self-Evolving-Agents","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EvoAgentX%2FAwesome-Self-Evolving-Agents/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EvoAgentX%2FAwesome-Self-Evolving-Agents/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EvoAgentX%2FAwesome-Self-Evolving-Agents/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/EvoAgentX","download_url":"https://codeload.github.com/EvoAgentX/Awesome-Self-Evolving-Agents/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/EvoAgentX%2FAwesome-Self-Evolving-Agents/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":36748743,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-08-06T04:43:03.162Z","status":"online","status_checked_at":"2026-08-17T02:00:06.698Z","response_time":86,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"created_at":"2025-11-06T22:02:56.417Z","updated_at":"2026-08-17T20:00:26.540Z","primary_language":null,"list_of_lists":false,"displayable":true,"categories":["1. Single-Agent Optimisation","3. Domain-Specific Optimisation","4. Evaluation","2. Multi-Agent Optimisation","Open-Source Framework"],"sub_categories":["1.1 🤖 LLM Behaviour Optimisation","1.4 🧰 Tool Optimization","3.2 💻 Programming","1.2 💬 Prompt Optimisation","4.1 📈 Benchmark-Based Evaluation","3.1 🧬 Biomedicine","1.3 🧠 Memory Optimization","2.2 🚀 MAS Optimisation","3.4 💰📚 Financial and Legal Research","4.3 🛡 Safety, Alignment, and Robustness for Lifelong / Self-Evolving Agents","4.2 ⚖️ LLM-Based Evaluation","1.5 🧰 Unified Optimization","3.5 🧩 Other Domain-Specific Optimisation","2.1 ⚙️ Automatic Multi-Agent Construction","3.3 Scientific Research"],"readme":"\u003c!-- \u003ch1 align=\"center\"\u003e\n  \u003cstrong\u003eA Comprehensive Survey of Self-Evolving AI Agents\u003cbr\u003eA New Paradigm Bridging Foundation Models and Lifelong Agentic Systems\u003c/strong\u003e\n\u003c/h1\u003e --\u003e\n\n\u003ch1 align=\"center\"\u003e\n  \u003cstrong\u003eAwesome-Self-Evolving-Agents\u003c/strong\u003e\n\u003c/h1\u003e\n\u003cdiv align=\"center\"\u003e\n\n\n[![Awesome](https://awesome.re/badge.svg?logo=stylelint)](https://awesome.re)\n[![arXiv](https://img.shields.io/badge/Arxiv-Self_Evolving_AI_Agents-b31b1b.svg?logo=arXiv)](https://arxiv.org/abs/2508.07407)\n[![Contribution Welcome](https://img.shields.io/badge/Contributions-welcome-Green?logo=mercadopago\u0026logoColor=white)](https://github.com/EvoAgentX/Awesome-Self-Evolving-Agents/pulls)\n[![GitHub star chart](https://img.shields.io/github/stars/EvoAgentX/Awesome-Self-Evolving-Agents?style=social)](https://star-history.com/#EvoAgentX/Awesome-Self-Evolving-Agents)\n[![License](https://img.shields.io/badge/License-MIT-yellow.svg?)](LICENSE)\n\n\u003ch3 align=\"center\"\u003e\n  \u003cstrong\u003e🤖 We're still cooking — Stay tuned! 🤖\u003cbr\u003e⭐ Give us a star if you like it! ⭐\u003c/strong\u003e\n\u003c/h3\u003e\n\n\u003cimg src=\"./assets/evolve-tree.jpg\" alt=\"Evolve Tree\"\u003e\n\u003cbr\u003e\n\u003cem\u003eFigure: A visual taxonomy of AI agent evolution and optimisation techniques, categorised into three major directions: single-agent optimisation, multi-agent optimisation, and domain-specific optimisation. The tree structure illustrates the development of these approaches from 2023 to 2025, including representative methods within each branch.\u003c/em\u003e\n\n\u003c/div\u003e\n\n## AI Agents Development Path\n\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"./assets/evolve-path.png\" alt=\"Development Path\", width=\"500\"\u003e\n\u003c/p\u003e\n\n## Conceptual Framework of the Self-Evolving AI Agents\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"./assets/evolve-framework.png\" alt=\"Conceptual Framework\", width=\"500\"\u003e\n\u003c/p\u003e\n\n## Open-Source Framework\n- (*EMNLP'25 Demo*) **EvoAgentX**: An Automated Framework for Evolving Agentic Workflows\n [[💻 Code](https://github.com/EvoAgentX/EvoAgentX)] [[📝 Paper](https://arxiv.org/abs/2507.03616)]\n- (*Arxiv'25*) MASLab: A Unified and Comprehensive Codebase for LLM-based Multi-Agent Systems [[📝 Paper](https://arxiv.org/abs/2505.16988)] [[💻 Code](https://github.com/MASWorks/MASLab)]\n\n\n## 1. Single-Agent Optimisation \n\n### 1.1 🤖 LLM Behaviour Optimisation \n\n#### 1.1.1 📌 Training-Based Behaviour Optimisation \n\n##### (1) 🔧 Supervised Fine-Tuning Approaches \n- (*ICLR'24*) ToRA: A tool-integrated reasoning agent for mathematical problem solving [[📝 Paper](https://arxiv.org/abs/2309.17452)] [[💻 Code](https://github.com/microsoft/ToRA)]\n- (*NeurIPS'22*) STaR : Bootstrapping reasoning with reasoning [[📝 Paper](https://arxiv.org/abs/2203.14465)] [[💻 Code](https://github.com/ezelikman/STaR)]\n- (*Arxiv'24*) NExT: Teaching large language models to reason about code execution [[📝 Paper](https://arxiv.org/abs/2404.14662)]\n- (*EMNLP'24*) MuMath-Code: Combining Tool-Use Large Language Models with Multi-perspective Data Augmentation for Mathematical Reasoning [[📝 Paper](https://arxiv.org/abs/2405.07551)]\n- (*ICML'25*) MAS-GPT: Training LLMs to build LLM-based multi-agent systems [[📝 Paper](https://arxiv.org/abs/2503.03686)] [[💻 Code](https://github.com/MASWorks/MAS-GPT)]\n\n##### (2) 🔧 Reinforcement Learning Approaches \n- (*ICML'24*) Self-Rewarding Language Models [[📝 Paper](https://arxiv.org/abs/2401.10020)] [[💻 Code](https://github.com/lucidrains/self-rewarding-lm-pytorch)]\n- (*Arxiv'24*) Tulu 3: Pushing Frontiers in Open Language Model Post-Training [[📝 Paper](https://arxiv.org/abs/2411.15124)] [[💻 Code](https://github.com/allenai/open-instruct)]\n- (*EMNLP'24*) Learning Planning-based Reasoning by Trajectories Collection and Process Reward Synthesizing [[📝 Paper](https://arxiv.org/abs/2402.00658)] [[💻 Code](https://github.com/SparkJiao/dpo-trajectory-reasoning)]\n- (*Arxiv'24*) Agent Q: Advanced Reasoning and Learning for Autonomous AI Agents [[📝 Paper](https://arxiv.org/abs/2408.07199)]\n- (*Arxiv'24*) DeepSeek-Prover: Advancing Theorem Proving in LLMs through Large-Scale Synthetic Data [[📝 Paper](https://arxiv.org/abs/2405.14333)]\n- (*ICML'25*) Diving into Self-Evolving Training for Multimodal Reasoning [[📝 Paper](https://arxiv.org/abs/2412.17451)] [[💻 Code](https://github.com/hkust-nlp/mstar)]\n- (*Arxiv'25*) Absolute Zero: Reinforced Self-play Reasoning with Zero Data [[📝 Paper](https://arxiv.org/abs/2505.03335)]\n- (*Arxiv'25*) R-Zero: Self-Evolving Reasoning LLM from Zero Data [[📝 Paper](https://arxiv.org/abs/2508.05004)] [[💻 Code](https://github.com/Chengsong-Huang/R-Zero)]\n- (*Arxiv'25*) SPIRAL: Self-Play on Zero-Sum Games Incentivizes Reasoning via Multi-Agent Multi-Turn Reinforcement Learning [[📝 Paper](https://arxiv.org/abs/2506.24119)] [[💻 Code](https://github.com/spiral-rl/spiral)]\n- (*Arxiv'25*) DistFlow: A Fully Distributed RL Framework for Scalable and Efficient LLM Post-Training [[📝 Paper](https://arxiv.org/abs/2507.13833)] [[💻 Code](https://github.com/sii-research/siiRL)]\n- (*Arxiv'25*) Vision-Zero: Scalable VLM Self-Improvement via Strategic Gamified Self-Play [[📝 Paper](https://www.arxiv.org/abs/2509.25541)] [[💻 Code](https://github.com/wangqinsi1/Vision-Zero)]\n- (*Arxiv'25*) Parallel-R1: Towards Parallel Thinking via Reinforcement Learning [[📝 Paper](https://arxiv.org/abs/2509.07980)] [[💻 Code](https://github.com/zhengkid/Parallel-R1)]\n- (*Arxiv'25*) SSRL: Self-Search Reinforcement Learning [[📝 Paper](https://arxiv.org/abs/2508.10874)] [[💻 Code](https://github.com/TsinghuaC3I/SSRL)]\n- (*Arxiv'25*) SeRL: Self-Play Reinforcement Learning for Large Language Models with Limited Data [[📝 Paper](https://arxiv.org/abs/2505.20347)] [[💻 Code](https://github.com/wantbook-book/SeRL)]\n\n\n#### 1.1.2 📌 Test-Time Behaviour Optimisation \n\n##### (1) 🔧 Feedback-Based Approaches \n- (*ICLR'23*) CodeT: Code Generation with Generated Tests [[📝 Paper](https://arxiv.org/abs/2207.10397)] [[💻 Code](https://github.com/microsoft/CodeT)]\n- (*ICML'23*) LEVER: Learning to Verify Language-to-Code Generation with Execution [[📝 Paper](https://arxiv.org/abs/2302.08468)] [[💻 Code](https://github.com/niansong1996/lever)]\n- (*ESEC/FSE'23*) Baldur: Whole-Proof Generation and Repair with Large Language Models [[📝 Paper](https://arxiv.org/abs/2303.04910)]\n- (*ACL'24*) Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations [[📝 Paper](https://arxiv.org/abs/2312.08935)]\n- (*EMNLP'24*) Learning Planning-based Reasoning by Trajectories Collection and Process Reward Synthesizing [[📝 Paper](https://arxiv.org/abs/2402.00658)] [[💻 Code](https://github.com/SparkJiao/dpo-trajectory-reasoning)]\n- (*Arxiv'24*) Skywork-Reward: Bag of Tricks for Reward Modeling in LLMs [[📝 Paper](https://arxiv.org/abs/2410.18451)]\n- (*ICLR'25*) Rewarding Progress: Scaling Automated Process Verifiers for LLM Reasoning [[📝 Paper](https://arxiv.org/abs/2410.08146)] \n- (*Arxiv'25*) Skywork-Reward-V2: Scaling Preference Data Curation via Human-AI Synergy [[📝 Paper](https://arxiv.org/abs/2507.01352)] [[💻 Code](https://github.com/SkyworkAI/Skywork-Reward-V2)]\n\n\n##### (2) 🔧 Search-Based Approaches \n- (*ICLR'23*) Self-consistency improves chain of thought reasoning in language models [[📝 Paper](https://arxiv.org/abs/2203.11171)]\n- (*ACL'23*) Solving Math Word Problems via Cooperative Reasoning induced Language Models [[📝 Paper](https://arxiv.org/abs/2210.16257)] [[💻 Code](https://github.com/TianHongZXY/CoRe)]\n- (*NeurIPS'23*) Tree of thoughts: Deliberate problem solving with large language models [[📝 Paper](https://arxiv.org/abs/2305.10601)] [[💻 Code](https://github.com/princeton-nlp/tree-of-thought-llm)]\n- (*NeurIPS'24*) Buffer of Thoughts: Thought-Augmented Reasoning with Large Language Models [[📝 Paper](https://arxiv.org/abs/2406.04271)] [[💻 Code](https://github.com/YangLing0818/buffer-of-thought-llm)]\n- (*COLM'24*) Deductive Beam Search: Decoding Deducible Rationale for Chain-of-Thought Reasoning [[📝 Paper](https://arxiv.org/abs/2401.17686)] [[💻 Code](https://github.com/OSU-NLP-Group/Deductive-Beam-Search)]\n- (*AAAI'24*) Graph of thoughts: Solving elaborate problems with large language models [[📝 Paper](https://arxiv.org/abs/2308.09687)] [[💻 Code](https://github.com/spcl/graph-of-thoughts)]\n- (*ICML'25*) Forest-of-Thought: Scaling Test-Time Compute for Enhancing LLM Reasoning [[📝 Paper](https://arxiv.org/abs/2412.09078)] [[💻 Code](https://github.com/iamhankai/Forest-of-Thought)]\n- (*NeurIPS'24*) AIDE: An Automatic Data Science Agent [[📝 Paper](https://arxiv.org/abs/2502.13138)] [[💻 Code](https://github.com/WecoAI/aideml)]\n\n##### （3）🔧 Reasoning-Based Approaches \n- (*EMNLP’25*) START: Self‑taught Reasoner with Tools [[📝 Paper](https://arxiv.org/abs/2503.04625)]\n- (*ArXiv’25*) CoRT: Code‑integrated Reasoning within Thinking [[📝 Paper](https://arxiv.org/abs/2506.09820)] [[💻 Code](https://github.com/ChengpengLi1003/CoRT)]\n\n### 1.2 💬 Prompt Optimisation\n\n#### 1.2.1 📌 Edit-Based Prompt Optimisation \n- (*EMNLP'22*) GPS: Genetic Prompt Search for Efficient Few-shot Learning [[📝 Paper](https://arxiv.org/abs/2210.17041)] [[💻 Code](https://github.com/hwxu20/GPS)]\n- (*EACL'23*) GrIPS: Gradient-free, Edit-based Instruction Search for Prompting Large Language Models [[📝 Paper](https://arxiv.org/abs/2203.07281)] [[💻 Code](https://github.com/archiki/GrIPS)]\n- (*ICLR'23*) TEMPERA: Test-Time Prompting via Reinforcement Learning [[📝 Paper](https://arxiv.org/abs/2211.11890)] [[💻 Code](https://github.com/tianjunz/TEMPERA)]\n- (*ACL'24*) Plum: Prompt Learning using Metaheuristic [[📝 Paper](https://arxiv.org/abs/2311.08364)] [[💻 Code](https://github.com/research4pan/Plum)]\n\n#### 1.2.2 📌 Evolutionary Prompt Optimisation \n- (*ICLR'24*) EvoPrompt: Connecting LLMs with Evolutionary Algorithms Yields Powerful Prompt Optimizers [[📝 Paper](https://arxiv.org/abs/2309.08532)] [[💻 Code](https://github.com/beeevita/EvoPrompt)]\n- (*ICML'24*) Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution [[📝 Paper](https://arxiv.org/abs/2309.16797)]\n- (*Arxiv'25*) GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning [[📝 Paper](https://arxiv.org/abs/2507.19457)]\n\n#### 1.2.3 📌 Generative Prompt Optimisation \n- (*ICLR'23*) Large Language Models Are Human-Level Prompt Engineers [[📝 Paper](https://arxiv.org/abs/2211.01910)] [[💻 Code](https://github.com/keirp/automatic_prompt_engineer)]\n- (*ICLR'24*) PromptAgent: Strategic Planning with Language Models Enables Expert-level Prompt Optimization [[📝 Paper](https://arxiv.org/abs/2310.16427)] [[💻 Code](https://github.com/XinyuanWangCS/PromptAgent)]\n- (*ICLR'24*) Large Language Models as Optimizers [[📝 Paper](https://arxiv.org/abs/2309.03409)] [[💻 Code](https://github.com/google-deepmind/opro)]\n- (*ICLR'24*) Retroformer: Retrospective Large Language Agents with Policy Gradient Optimization [[📝 Paper](https://arxiv.org/abs/2308.02151)] [[💻 Code](https://github.com/weirayao/Retroformer)]\n- (*EMNLP'24*) Optimizing Instructions and Demonstrations for Multi-Stage Language Model Programs [[📝 Paper](https://arxiv.org/abs/2406.11695)] [[💻 Code](https://github.com/stanfordnlp/dspy)]\n- (*Arxiv'24*) Prompt Optimization with Human Feedback [[📝 Paper](https://arxiv.org/abs/2405.17346)] [[💻 Code](https://github.com/xqlin98/APOHF)]\n- (*Arxiv'24*) StraGo: Harnessing Strategic Guidance for Prompt Optimization [[📝 Paper](https://arxiv.org/abs/2410.08601)]\n- (*Arxiv'25*) Self-Supervised Prompt Optimization [[📝 Paper](https://arxiv.org/abs/2502.06855)] \n\n#### 1.2.4 📌 Text Gradient-Based Prompt Optimisation \n- (*EMNLP'23*) Automatic Prompt Optimization with \"Gradient Descent\" and Beam Search [[📝 Paper](https://arxiv.org/abs/2305.03495)] [[💻 Code](https://github.com/microsoft/LMOps/tree/main/prompt_optimization)]\n- (*Arxiv'24*) TextGrad: Automatic \"Differentiation\" via Text [[📝 Paper](https://arxiv.org/abs/2406.07496)] [[💻 Code](https://github.com/zou-group/textgrad)]\n- (*Arxiv'24*) How to Correctly do Semantic Backpropagation on Language-based Agentic Systems [[📝 Paper](https://arxiv.org/abs/2412.03624)] [[💻 Code](https://github.com/HishamAlyahya/semantic_backprop)]\n- (*Arxiv'24*) GRAD-SUM: Leveraging Gradient Summarization for Optimal Prompt Engineering [[📝 Paper](https://arxiv.org/abs/2407.12865)]\n- (*AAAI'25*) Unleashing the Potential of Large Language Models as Prompt Optimizers: Analogical Analysis with Gradient-based Model Optimizers [[📝 Paper](https://arxiv.org/abs/2402.17564)] [[💻 Code](https://github.com/RUCAIBox/GPO)]\n- (*ICML'25*) REVOLVE: Optimizing AI Systems by Tracking Response Evolution in Textual Optimization [[📝 Paper](https://arxiv.org/abs/2412.03092)] [[💻 Code](https://github.com/Peiyance/REVOLVE)]\n- (*Arxiv'25*) PersonaAgent: When Large Language Model Agents Meet Personalization at Test Time [[📝 Paper](https://arxiv.org/abs/2506.06254)]\n \n### 1.3 🧠 Memory Optimization\n- (*ICML'24*) A Human-Inspired Reading Agent with Gist Memory of Very Long Contexts [[📝 Paper](https://arxiv.org/abs/2402.09727)]\n- (*ICML'24*) Agent Workflow Memory [[📝 Paper](https://arxiv.org/abs/2409.07429)] \n- (*AAAI'24*) MemoryBank: Enhancing Large Language Models with Long-Term Memory [[📝 Paper](https://arxiv.org/abs/2305.10250)]\n- (*EMNLP'24*) GraphReader: Building graph-based agent to enhance long-context [[📝 Paper](https://arxiv.org/abs/2406.14550)]\n- (*Arxiv'24*) \"My agent understands me better\": Integrating Dynamic Human-like Memory Recall and Consolidation in LLM-Based Agents [[📝 Paper](https://arxiv.org/abs/2404.00573)] \n- (*ICLR'25*) Compress to Impress: Unleashing the Potential of Compressive Memory in Real-World Long-Term Conversations [[📝 Paper](https://arxiv.org/abs/2402.11975)]\n- (*ICLR'25*) Boosting knowledge intensive reasoning of llms via inference-time hybrid information [[📝 Paper](https://arxiv.org/abs/2410.08815)] [[💻 Code](https://github.com/icip-cas/StructRAG)]\n- (*ACL'25*) Improving factuality with explicit working memory [[📝 Paper](https://arxiv.org/abs/2412.18069)]\n- (*Arxiv'25*) A-MEM: Agentic Memory for LLM Agents [[📝 Paper](https://arxiv.org/abs/2502.12110)]\n- (*Arxiv'25*) Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory [[📝 Paper](https://arxiv.org/abs/2504.19413)]\n- (*Arxiv'25*) Memento: Fine‑tuning LLM Agents without Fine‑tuning LLMs [[📝 Paper](https://www.arxiv.org/abs/2508.16153)] [[💻 Code](https://github.com/Agent-on-the-Fly/Memento)]\n- (*Arxiv'25*) Memory-R1: Enhancing Large Language Model Agents to Manage and Utilize Memories via Reinforcement Learning [[📝 Paper](https://arxiv.org/abs/2508.19828)]\n- (*Arxiv'25*) Seeing, Listening, Remembering, and Reasoning: A Multimodal Agent with Long-Term Memory [[📝 Paper](https://arxiv.org/abs/2508.09736)] [[💻 Code](https://github.com/bytedance-seed/m3-agent)] \n- (*Arxiv'25*) PersonaAgent: When Large Language Model Agents Meet Personalization at Test Time [[📝 Paper](https://arxiv.org/abs/2506.06254)]  \n- (*Arxiv'25*) ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory [[📝 Paper](https://arxiv.org/abs/2509.25140)]\n\n### 1.4 🧰 Tool Optimization\n\n#### 1.4.1 📌 Training-Based Tool Optimisation\n\n##### (1) Supervised Fine-Tuning for Tool Optimisation \n- (*NeurIPS'23*) GPT4Tools: Teaching Large Language Model to Use Tools via Self-instruction [[📝 Paper](https://arxiv.org/abs/2305.18752)] [[💻 Code](https://github.com/AILab-CVC/GPT4Tools)] \n- (*ICLR'24*) ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs [[📝 Paper](https://arxiv.org/abs/2307.16789)] [[💻 Code](https://github.com/OpenBMB/ToolBench)]\n- (*ACL'24*) LLMs in the Imaginarium: Tool Learning through Simulated Trial and Error [[📝 Paper](https://arxiv.org/abs/2403.04746)] [[💻 Code](https://github.com/microsoft/simulated-trial-and-error)]\n- (*AAAI'24*) Confucius: Iterative tool learning from introspection feedback by easy-to-difficult curriculum [[📝 Paper](https://arxiv.org/abs/2308.14034)] [[💻 Code](https://github.com/shizhl/Confucius)]\n- (*ICLR'25*) Learning Evolving Tools for Large Language Models [[📝 Paper](https://arxiv.org/abs/2410.06617)] [[💻 Code](https://github.com/Chen-GX/ToolEVO)] \n- (*ICLR'25*) Facilitating Multi-turn Function Calling for LLMs via Compositional Instruction Tuning [[📝 Paper](https://arxiv.org/abs/2410.12952)] [[💻 Code](https://github.com/PKU-Baichuan-MLSystemLab/BUTTON)] \n- (*ICLR'25*) Multi-modal Agent Tuning: Building a VLM-Driven Agent for Efficient Tool Usage [[📝 Paper](https://arxiv.org/abs/2412.15606)] [[💻 Code](https://github.com/mat-agent/MAT-Agent)] \n- (*Arxiv'25*) Magnet: Multi-turn Tool-use Data Synthesis and Distillation via Graph Translation [[📝 Paper](https://arxiv.org/abs/2503.07826)]\n- (*ICML'25*) Adapting While Learning: Grounding LLMs for Scientific Problems with Intelligent Tool Usage Adaptation [[📝 Paper](https://arxiv.org/abs/2411.00412)] [[💻 Code](https://github.com/Rose-STL-Lab/Adapting-While-Learning)]\n\n\n##### (2) Reinforcement Learning for Tool Optimisation \n- (*Arxiv'25*) ReTool: Reinforcement Learning for Strategic Tool Use in LLMs [[📝 Paper](https://arxiv.org/abs/2504.11536)] [[💻 Code](https://github.com/ReTool-RL/ReTool)] \n- (*Arxiv'25*) ToolRL: Reward is All Tool Learning Needs [[📝 Paper](https://arxiv.org/abs/2504.13958)] [[💻 Code](https://github.com/qiancheng0/ToolRL)]\n- (*Arxiv'25*) Nemotron-Research-Tool-N1: Exploring Tool-Using Language Models with Reinforced Reasoning [[📝 Paper](https://arxiv.org/abs/2505.00024)] [[💻 Code](https://github.com/NVlabs/Tool-N1)] \n- (*Arxiv'25*) Synthetic Data Generation \u0026 Multi-Step RL for Reasoning \u0026 Tool Use [[📝 Paper](https://arxiv.org/abs/2504.04736)] \n- (*Arxiv'25*) Iterative Tool Usage Exploration for Multimodal Agents via Step-wise Preference Tuning [[📝 Paper](https://arxiv.org/abs/2504.21561)] [[💻 Code](https://github.com/SPORT-Agents/SPORT-Agents)]  \n- (*Arxiv'25*) Tool-Star: Empowering LLM-Brained Multi-Tool Reasoner via Reinforcement Learning [[📝 Paper](https://arxiv.org/abs/2505.16410)] [[💻 Code](https://github.com/RUC-NLPIR/Tool-Star)] \n- (*Arxiv'25*) Agentic Reinforced Policy Optimization [[📝 Paper](https://arxiv.org/abs/2507.19849)] [[💻 Code](https://github.com/RUC-NLPIR/ARPO)]\n- (*Arxiv'25*) AutoTIR: Autonomous Tools Integrated Reasoning via Reinforcement Learning [[📝 Paper](https://arxiv.org/abs/2507.21836)] [[💻 Code](https://github.com/weiyifan1023/AutoTIR)]\n\n\n#### 1.4.2 📌 Inference-Time Tool Optimisation \n\n##### (1) Prompt-Based Optimisation \n- (*NAACL'25*) EASYTOOL: Enhancing LLM-based Agents with Concise Tool Instruction [[📝 Paper](https://arxiv.org/abs/2401.06201)] [[💻 Code](https://github.com/microsoft/JARVIS/tree/main/easytool)]\n- (*ICLR'25*) From Exploration to Mastery: Enabling LLMs to Master Tools via Self-Driven Interactions [[📝 Paper](https://arxiv.org/abs/2410.08197)] [[💻 Code](https://github.com/quchangle1/DRAFT)] \n- (*ACL'25*) Zero-shot Tool Instruction Optimization for LLM Agents via Tool Play [[📝 Paper](https://arxiv.org/abs/2503.14432)] [[💻 Code](https://github.com/wfangtw/play2prompt)] \n\n##### (2) Reasoning-Based Optimisation \n- (*ICLR'24*) ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs [[📝 Paper](https://arxiv.org/abs/2307.16789)] [[💻 Code](https://github.com/beijixiong1/ToolLLM)]\n- (*ICLR'24*) ToolChain*: Efficient Action Space Navigation in Large Language Models with A* Search [[📝 Paper](https://arxiv.org/abs/2310.13227)] \n- (*ICLR'25*) Tool-Planner: Task Planning with Clusters across Multiple Tools [[📝 Paper](https://arxiv.org/abs/2406.03807)] [[💻 Code](https://github.com/OceannTwT/Tool-Planner)] \n- (*Arxiv'25*) MCP-Zero: Active Tool Discovery for Autonomous LLM Agents [[📝 Paper](https://arxiv.org/abs/2506.01056)][[💻 Code](https://github.com/xfey/MCP-Zero)]\n\n\n#### 1.4.3 📌 Tool Functionality Optimisation \n- (*EMNLP'23*) CREATOR : Tool creation for disentangling abstract and concrete reasoning of large language model [[📝 Paper](https://arxiv.org/abs/2305.14318)] [[💻 Code](https://github.com/qiancheng0/CREATOR)]\n- (*ICML'24*) Offline Training of Language Model Agents with Functions as Learnable Weights [[📝 Paper](https://arxiv.org/abs/2402.11359)]\n- (*CVPR'24*) CLOVA: A Closed-Loop Visual Assistant with Tool Usage and Update [[📝 Paper](https://arxiv.org/abs/2312.10908)] [[💻 Code](https://github.com/clova-tool/CLOVA-tool)]\n- (*Arxiv'25*) Alita: Generalist Agent Enabling Scalable Agentic Reasoning with Minimal Predefinition and Maximal Self-Evolution [[📝 Paper](https://arxiv.org/abs/2505.20286)] [[💻 Code](https://github.com/CharlesQ9/Alita)]\n\n\n### 1.5 🧰 Unified Optimization\n- (*Arxiv'25*) Building Self-Evolving Agents via Experience-Driven Lifelong Learning: A Framework and Benchmark [[📝 Paper](https://arxiv.org/abs/2508.19005)] [[💻 Code](https://github.com/ECNU-ICALK/ELL-StuLife)]\n- (*Arxiv'25*) EvoAgent: Self-evolving Agent with Continual World Model for Long-Horizon Tasks [[📝 Paper](https://arxiv.org/abs/2502.05907)]\n\n\n## 2. Multi-Agent Optimisation \n\n### 2.1 ⚙️ Automatic Multi-Agent Construction\n- （*ICML'25*) MetaAgent: Automatically Constructing Multi-Agent Systems Based on Finite State Machines [[📝 Paper](https://arxiv.org/abs/2507.22606)] [[💻 Code](https://github.com/SaFoLab-WISC/MetaAgent/)]\n\n### 2.2 🚀 MAS Optimisation \n- (*Arxiv'26*) CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery [[📝 Paper](https://arxiv.org/abs/2604.01658)] [[💻 Code](https://github.com/Human-Agent-Society/CORAL)]\n- (*Arxiv' 25*) R\u0026D-Agent: Automating Data-Driven AI Solution Building Through LLM-Powered Automated Research, Development, and Evolution [[📝 Paper](https://arxiv.org/abs/2505.14738)] [[💻 Code](https://github.com/microsoft/RD-Agent)]\n- (*ICML'25*) Multi-Agent Architecture Search via Agentic Supernet [[📝 Paper](https://doi.org/10.48550/arXiv.2502.04180)] [[💻Code](https://github.com/bingreeky/MaAS)]\n- (*ICML'25*) MA-LoT: Multi-Agent Lean-based Long Chain-of-Thought Reasoning enhances Formal Theorem Proving [[📝 Paper](https://arxiv.org/abs/2503.03205)]\n- (*ICLR'25*) AFlow: Automating Agentic Workflow Generation [[📝 Paper](https://arxiv.org/abs/2410.10762)] [[💻 Code](https://github.com/geekan/MetaGPT/tree/main/examples/aflow)]\n- (*ICLR'25*) WorkflowLLM: Enhancing Workflow Orchestration Capability of Large Language Models [[📝 Paper](https://openreview.net/forum?id=3Hy00Wvabi)]\n- (*ICLR'25*) Flow: Modularized Agentic Workflow Automation [[📝 Paper](https://openreview.net/forum?id=sLKDbuyq99)]\n- (*ICLR'25*) Automated Design of Agentic Systems [[📝 Paper](https://arxiv.org/abs/2408.08435)] [[💻 Code](https://github.com/ShengranHu/ADAS)]\n- (*Arxiv'25*) FlowReasoner: Reinforcing Query-Level Meta-Agents [[📝 Paper](https://doi.org/10.48550/arXiv.2504.15257)]\n- (*Arxiv'25*) AgentNet: Decentralized Evolutionary Coordination for LLM-Based Multi-Agent Systems [[📝 Paper](https://arxiv.org/abs/2504.00587)]\n- (*Arxiv'25*) MAS-GPT: Training LLMs to Build LLM-Based Multi-Agent Systems [[📝 Paper](https://arxiv.org/abs/2503.03686)]\n- (*Arxiv'25*) FlowAgent: Achieving Compliance and Flexibility for Workflow Agents [[📝 Paper](https://doi.org/10.48550/arXiv.2502.14345)]\n- (*Arxiv'25*) ScoreFlow: Mastering LLM Agent Workflows via Score-Based Preference Optimization [[📝 Paper](https://arxiv.org/abs/2502.04306)] [[💻 Code](https://github.com/Gen-Verse/ScoreFlow)]\n- (*Arxiv'25*) Multi-Agent Design: Optimizing Agents with Better Prompts and Topologies [[📝 Paper](https://arxiv.org/abs/2502.02533)]\n- (*Arxiv'25*) MAS-ZERO: Designing Multi-Agent Systems with Zero Supervision [[📝 Paper](https://arxiv.org/abs/2505.14996)]\n- (*Arxiv'25*) MermaidFlow: Redefining Agentic Workflow Generation via Safety-Constrained Evolutionary Programming [[📝 Paper](https://arxiv.org/abs/2505.22967)]\n- (*ICML'24*) GPTSwarm: Language Agents as Optimizable Graphs [[📝 Paper](https://arxiv.org/abs/2402.16823)] [[Code](https://github.com/metauto-ai/gptswarm)]\n- (*ICLR'24*) DSPy: Compiling Declarative Language Model Calls into State-of-the-Art Pipelines [[📝 Paper](https://openreview.net/forum?id=sY5N0zY5Od)] [[💻 Code](https://github.com/stanfordnlp/dspy)]\n- (*ICLR'24*) AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors [[📝 Paper](https://openreview.net/forum?id=EHg5GDnyq1)] [[💻 Code](https://github.com/OpenBMB/AgentVerse)]\n- (*ICLR'24*) MetaGPT: Meta Programming for a Multi-Agent Collaborative Framework [[📝 Paper](https://openreview.net/forum?id=VtmBAGCN7o)] [[💻 Code](https://github.com/geekan/MetaGPT)]\n- (*COLM'24*) A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaboration [[📝 Paper](https://openreview.net/forum?id=XII0Wp1XA9)]\n- (*COLM'24*) AutoGen: Enabling next-Gen LLM Applications via Multi-Agent Conversations [[📝 Paper](https://openreview.net/forum?id=BAakY1hNKS)] [[💻 Code](https://github.com/microsoft/autogen)]\n- (*Arxiv'24*) G-Designer: Architecting Multi-Agent Communication Topologies via Graph Neural Networks [[📝 Paper](https://doi.org/10.48550/arXiv.2410.11782)]\n- (*Arxiv'24*) AutoFlow: Automated Workflow Generation for Large Language Model Agents [[📝 Paper](https://arxiv.org/abs/2407.12821)] [[💻 Code](https://github.com/agiresearch/AutoFlow)]\n- (*Arxiv'24*) Symbolic Learning Enables Self-Evolving Agents [[📝 Paper](https://arxiv.org/abs/2406.18532)] [[💻 Code](https://github.com/aiwaves-cn/agents)]\n- (*Arxiv'24*) Adaptive In-Conversation Team Building for Language Model Agents [[📝 Paper](https://arxiv.org/abs/2405.19425)]\n- (*ICLR'25*) Self-Evolving Multi-Agent Collaboration Networks for Software Development [[📝 Paper](https://openreview.net/forum?id=4R71pdPBZp)] [[💻 Code](https://github.com/yuzhu-cai/rSDE-Bench)] \n- (*Arxiv'25*) Chain‑of‑Agents: End‑to‑End Agent Foundation Models via Multi‑Agent Distillation and Agentic RL [[📝 Paper](https://arxiv.org/abs/2508.13167)] [[💻 Code](https://github.com/OPPO-PersonalAI/Agent_Foundation_Models)]\n- (*Arxiv’25*) Agent KB: Leveraging Cross‑Domain Experience for Agentic Problem Solving [[📝 Paper](https://arxiv.org/abs/2507.06229)] [[💻 Code](https://github.com/OPPO-PersonalAI/Agent-KB)]\n\n\n## 3. Domain-Specific Optimisation \n\n### 3.1 🧬 Biomedicine \n\n#### 3.1.1 📌 Medical Diagnosis \n\n- (*EMNLP'24*) MMedAgent: Learning to Use Medical Tools with Multi-modal Agent [[📝 Paper](https://arxiv.org/abs/2407.02483)] [[💻 Code](https://github.com/Wangyixinxin/MMedAgent)]\n- (*NeurIPS'24*) MDAgents: An Adaptive Collaboration of LLMs for Medical Decision-Making [[📝 Paper](https://arxiv.org/abs/2404.15155)] [[💻 Code](https://github.com/mitmedialab/MDAgents)]\n- (*Arxiv'25*) HealthFlow: A Self-Evolving AI Agent with Meta Planning for Autonomous Healthcare Research [[📝 Paper](https://arxiv.org/pdf/2508.02621)][[💻 Code](https://github.com/yhzhu99/HealthFlow)]\n- (*Arxiv'25*) STELLA: Self-Evolving LLM Agent for Biomedical Research [[📝 Paper](https://arxiv.org/abs/2507.02004)][[💻 Code](https://github.com/zaixizhang/STELLA)]\n- (*MICCAI'25*) MedAgentSim: Self-Evolving Multi-Agent Simulations for Realistic Clinical Interactions [[📝 Paper](https://arxiv.org/pdf/2503.22678)] [[💻 Code](https://github.com/MAXNORM8650/MedAgentSim)]\n- (*Arxiv'25*) PathFinder: A Multi-Modal Multi-Agent System\nfor Medical Diagnostic Decision-Making Applied to Histopathology \u003cbr\u003e [[📝 Paper](https://arxiv.org/pdf/2502.08916)]\n- (*Arxiv'25*) MDTeamGPT: A Self-Evolving LLM-based Multi-Agent Framework for Multi-Disciplinary Team Medical Consultation \u003cbr\u003e [[📝 Paper](https://arxiv.org/abs/2503.13856)] [[💻 Code](https://github.com/KaiChenNJ/MDTeamGPT)]\n- (*Arxiv'25*) MedAgent-Pro: Towards Evidence-based Multi-modal\nMedical Diagnosis via Reasoning Agentic Workflow \u003cbr\u003e [[📝 Paper](https://arxiv.org/pdf/2503.18968)] [[💻 Code](https://github.com/jinlab-imvr/MedAgent-Pro)]\n- (*Arxiv'25*) Structural Entropy Guided Agent for Detecting and Repairing Knowledge Deficiencies in LLMs [[📝 Paper](https://arxiv.org/abs/2505.07184)] [[💻 Code](https://github.com/weiyifan1023/senator)]\n\n\n#### 3.1.2 📌  Molecular Discovery \n\n- (*ACS omega'24*) CACTUS: Chemistry Agent Connecting Tool-Usage to Science [[📝 Paper](https://arxiv.org/abs/2405.00972)] [[💻 Code](https://github.com/pnnl/cactus)]\n- (*NMI'24*) ChemCrow: Augmenting large language models with chemistry tools [[📝 Paper](https://arxiv.org/abs/2304.05376)] [[💻 Code](https://github.com/ur-whitelab/chemcrow-public)]\n- (*ICLR'25*) ChemAgent: Self-updating Library in Large Language Models Improves Chemical Reasoning[[📝 Paper](https://arxiv.org/abs/2501.06590)] [[💻 Code](https://github.com/gersteinlab/ChemAgent)]\n- (*ICLR'25*) OSDA Agent: Leveraging Large Language Models for De Novo Design of Organic Structure Directing Agents [[📝 Paper](https://openreview.net/forum?id=9YNyiCJE3k)]\n- (*Arxiv'25*) DrugAgent: Automating AI-aided Drug Discovery Programming through\nLLM Multi-Agent Collaboration [[📝 Paper](https://arxiv.org/pdf/2411.15692)]\n- (*Arxiv'25*) LIDDIA: Language-based Intelligent Drug Discovery Agent [[📝 Paper](https://arxiv.org/abs/2502.13959)]\n- (*Arxiv'25*) GenoMAS: A Multi-Agent Framework for Scientific Discovery via Code-Driven Gene Expression Analysis \u003cbr\u003e [[📝 Paper](https://arxiv.org/abs/2507.21035)] [[💻 Code](https://github.com/Liu-Hy/GenoMAS)]\n\n### 3.2 💻 Programming\n\n#### 3.2.1 📌 Code Refinement \n\n- (*Arxiv'23*) AgentCoder: Multi-Agent-based Code Generation with Iterative Testing and Optimisation [[📝 Paper](https://arxiv.org/abs/2312.13010)] [[💻 Code](https://github.com/huangd1999/AgentCoder)]\n- (*Arxiv'23*) Self-Refine: Iterative Refinement with Self-Feedback [[📝 Paper](https://arxiv.org/abs/2303.17651)] [[💻 Code](https://github.com/madaan/self-refine)]\n- (*EMNLP'24*) CodeAgent: Autonomous Communicative Agents for Code Review [[📝 Paper](https://aclanthology.org/2024.emnlp-main.632/)] [[💻 Code](https://github.com/Daniel4SE/codeagent)]\n- (*ICLR'25*) OpenHands: An Open Platform for AI Software Developers as Generalist Agents [[📝 Paper](https://arxiv.org/abs/2407.16741)] [[💻 Code](https://github.com/All-Hands-AI/OpenHands)]\n- (*Arxiv'25*) CodeCoR: An LLM-Based Self-Reflective Multi-Agent Framework for Code Generation [[📝 Paper](https://arxiv.org/abs/2501.07811)]\n- (Arxiv’25) AlphaEvolve: A coding agent for scientific and algorithmic discovery [[📝 Paper](https://arxiv.org/abs/2506.13131)]\n- (*Arxiv'25*) Darwin Gödel Machine: Open-Ended Evolution of Self-Improving Agents [[📝 Paper](https://arxiv.org/pdf/2505.22954)] [[💻 Code](https://github.com/jennyzzt/dgm)] \n- (*Software'25*) OpenEvolve: an open-source evolutionary coding agent [[📝 Instructions](https://huggingface.co/blog/codelion/openevolve)] [[💻 Code](https://github.com/codelion/openevolve)]\n- (*ICLR'25*) Self-Evolving Multi-Agent Collaboration Networks for Software Development [[📝 Paper](https://openreview.net/forum?id=4R71pdPBZp)] [[💻 Code](https://github.com/yuzhu-cai/rSDE-Bench)]\n- (*ASE'25*) Live-SWE-agent: Can Software Engineering Agents Self-Evolve on the Fly? [[📝 Paper](https://arxiv.org/abs/2511.13646)]\n\n#### 3.2.2 📌 Code Debugging\n\n- (*ACL'23*) Self-Edit: Fault-Aware Code Editor for Code Generation [[📝 Paper](https://arxiv.org/abs/2305.04087)]\n- (*ICLR'24*) Teaching Large Language Models to Self-Debug [[📝 Paper](https://arxiv.org/abs/2304.05128)]\n- (*ICA'24*) RGD: Multi-LLM based agent debugger via refinement and generation guidance. [[📝 Paper](https://arxiv.org/abs/2410.01242)]\n- (*Arxiv'25*) Large Language Model Guided Self-Debugging Code Generation [[📝 Paper](https://arxiv.org/abs/2502.02928)]\n\n\n### 3.3 Scientific Research \n- (*Arxiv’25*) PiFlow: Principle‑aware Scientific Discovery with Multi‑Agent Collaboration [[📝 Paper](https://arxiv.org/abs/2505.15047)] [[💻 Code](https://github.com/amair-lab/PiFlow)]\n\n\n### 3.4 💰📚 Financial and Legal Research \n\n#### 3.4.1 📌 Financial Decision-Making \n\n- (*Arxiv'25*) R\u0026D-Agent-Quant: A Multi-Agent Framework for Data-Centric Factors and Model Joint Optimization [[📝 Paper](https://arxiv.org/abs/2505.15155)] [[💻 Code](https://github.com/microsoft/RD-Agent)]\n- (*Arxiv'24*) FinRobot: an open-source ai agent platform for financial applications using large language\nmodels [[📝 Paper](https://arxiv.org/abs/2405.14767)] [[💻 Code](https://github.com/AI4Finance-Foundation/FinRobot)]\n- (*Arxiv'24*) PEER: Expertizing domain-specific tasks with a multi-agent framework and tuning methods [[📝 Paper](https://arxiv.org/abs/2407.06985)] [[💻 Code](https://github.com/agentuniverse-ai/agentUniverse)]\n- (*NeurIPS'25*) Fincon: A synthesized llm multi-agent system with conceptual verbal reinforcement for enhanced\nfinancial decision making [[📝 Paper](https://proceedings.neurips.cc/paper_files/paper/2024/file/f7ae4fe91d96f50abc2211f09b6a7e49-Paper-Conference.pdf)]  [[💻 Code](https://github.com/The-FinAI/FinCon)]\n\n\n#### 3.4.2 📌 Legal Reasoning\n- (*Arxiv'24*) LawLuo: A Multi-Agent Collaborative Framework for Multi-Round Chinese Legal Consultation [[📝 Paper](https://arxiv.org/abs/2407.16252)]\n- (*ICIC'24*) Legalgpt: Legal chain of thought for the legal large\nlanguage model multi-agent framework [[📝 Paper](https://dl.acm.org/doi/10.1007/978-981-97-5678-0_3)]\n- (*Arxiv'24*) LawGPT: A Chinese Legal Knowledge-Enhanced Large Language Model\n [[📝 Paper](https://arxiv.org/abs/2406.04614)] [[💻 Code](https://github.com/pengxiao-song/LaWGPT)]\n - (*ACL Findings'25*) AgentCourt: Simulating Court with Adversarial Evolvable Lawyer Agents [[📝 Paper](https://aclanthology.org/2025.findings-acl.304/)] [[💻 Code](https://github.com/relic-yuexi/AgentCourt)]\n\n\n### 3.5 🧩 Other Domain-Specific Optimisation\n- (*Arxiv'25*) Agents of Change: Self-Evolving LLM Agents for Strategic Planning [[📝 Paper](https://arxiv.org/abs/2506.04651)]\n- (*Arxiv'25*) EarthLink: A Self-Evolving AI Agent for Climate Science [[📝 Paper](https://arxiv.org/abs/2507.17311)] [[🖥️ System](https://earthlink.intern-ai.org.cn/)]\n- (*Arxiv'25*) SEAgent: Self-Evolving Computer Use Agent with Autonomous Learning from Experience [[📝 Paper](https://arxiv.org/abs/2508.04700)][[💻 Code](https://github.com/SunzeY/SEAgent)]\n\n\n## 4. Evaluation \n\n### 4.1 📈 Benchmark-Based Evaluation \n\n- (*NeurIPS'23*) OpenAGI: When LLM Meets Domain Experts [[📝 Paper](https://arxiv.org/abs/2304.04370)] [[💻 Code](https://github.com/agiresearch/OpenAGI)]\n- (*Arxiv'25*) Building Self-Evolving Agents via Experience-Driven Lifelong Learning: A Framework and Benchmark [[📝 Paper](https://arxiv.org/abs/2508.19005)]\n- (*Arxiv'25*) MLGym: A New Framework and Benchmark for Advancing AI Research Agents [[📝 Paper](https://arxiv.org/abs/2502.14499)] [[💻 Code](https://github.com/facebookresearch/MLGym)]\n- (*Arxiv'25*) X-MAS: Towards Building Multi-Agent Systems with Heterogeneous LLMs [[📝 Paper](https://arxiv.org/abs/2505.16997)] [[💻 Code](https://github.com/MASWorks/X-MAS)]\n\n#### 4.1.1 📌 Tool and API-Driven Agents \n- (*Arxiv'23*) On the Tool Manipulation Capability of Open-source Large Language Models [[📝 Paper](https://arxiv.org/abs/2305.16504)] [[💻 Code](https://github.com/sambanova/toolbench)]\n- (*EMNLP'23*) API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs [[📝 Paper](https://arxiv.org/abs/2304.08244)] [[💻 Code](https://github.com/AlibabaResearch/DAMO-ConvAI/tree/main/api-bank)]\n- (*NeurIPS'23*) ToolQA: A Dataset for LLM Question Answering with External Tools [[📝 Paper](https://arxiv.org/abs/2306.13304)] [[💻 Code](https://github.com/night-chen/ToolQA)]\n- (*ICLR'24*) MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Use\n [[📝 Paper](https://arxiv.org/abs/2310.03128)] [[💻 Code](https://github.com/HowieHwong/MetaTool)]\n- (*Arxiv'25*) Enhancing Open-Domain Task-Solving Capability of LLMs via Autonomous Tool Integration from GitHub [[📝 Paper](https://arxiv.org/abs/2312.17294)] [[💻 Code](https://github.com/OpenBMB/OpenAct)]\n- (*Arxiv'25*) LiveMCP-101: Stress Testing and Diagnosing MCP-enabled Agents on Challenging Queries [[📝 Paper](https://arxiv.org/abs/2508.15760)]\n\n#### 4.1.2 📌 Web Navigation and Browsing Agents \n- (*ICLR'24*) WebArena: A Realistic Web Environment for Building Autonomous Agents [[📝 Paper](https://arxiv.org/abs/2307.13854)] [[💻 Code](https://github.com/web-arena-x/webarena)]\n- (*Arxiv'25*) BrowseComp: A Simple Yet Challenging Benchmark for Browsing Agents [[📝 Paper](https://arxiv.org/abs/2504.12516)] [[💻 Code](https://github.com/openai/simple-evals)]\n- (*ACL'25*) WebWalker: Benchmarking LLMs in Web Traversal [[📝 Paper](https://arxiv.org/abs/2501.07572)] [[💻 Code](https://github.com/Alibaba-NLP/WebAgent)]\n\n#### 4.1.3 📌 Coding Agents \n- (*ICLR'24*) SWE-bench: Can Language Models Resolve Real-World GitHub Issues? [[📝 Paper](https://arxiv.org/abs/2310.06770)] [[💻 Code](https://github.com/SWE-bench/SWE-bench)]\n- (*ICLR'25*) Self-Evolving Multi-Agent Collaboration Networks for Software Development [[📝 Paper](https://openreview.net/forum?id=4R71pdPBZp)] [[💻 Code](https://github.com/yuzhu-cai/rSDE-Bench)] \n\n#### 4.1.4 Scientific Research Agents \n- (*Arxiv'25*) DataSciBench: An LLM Agent Benchmark for Data Science [[📝 Paper](https://arxiv.org/abs/2502.13897)] [[💻 Code](https://github.com/THUDM/DataSciBench)]\n\n#### 4.1.4 📌 Multi-Agent Collaboration and Generalists \n- (*ICLR'23*) GAIA: a benchmark for General AI Assistants [[📝 Paper](https://arxiv.org/abs/2311.12983)] [[💻 Code](https://huggingface.co/gaia-benchmark)]\n- (*ICLR'24*) AgentBench: Evaluating LLMs as Agents [[📝 Paper](https://arxiv.org/abs/2308.03688)] [[💻 Code](https://github.com/THUDM/AgentBench)]\n- (*Arxiv'25*) MultiAgentBench: Evaluating the Collaboration and Competition of LLM agents [[📝 Paper](https://arxiv.org/abs/2503.01935)] [[💻 Code](https://github.com/MultiagentBench/MARBLE)]\n- (*Arxiv'25*) Benchmarking LLMs' Swarm intelligence [[📝 Paper](https://arxiv.org/abs/2505.04364)] [[💻 Code](https://github.com/RUC-GSAI/YuLan-SwarmIntell)]\n\n#### 4.1.5 📌 GUI and Multimodal Environment Agents\n- (*ACL'24*) Mobile-Bench: An Evaluation Benchmark for LLM-based Mobile Agents [[📝 Paper](https://arxiv.org/abs/2407.00993)] [[💻 Code](https://github.com/XiaoMi/MobileBench)]\n- (*NeurIPS'24*) OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments [[📝 Paper](https://arxiv.org/abs/2404.07972)] [[💻 Code](https://github.com/xlang-ai/OSWorld)]\n- (*ICLR'25*) AndroidWorld: A Dynamic Benchmarking Environment for Autonomous Agents [[📝 Paper](https://arxiv.org/abs/2405.14573)] [[💻 Code](https://github.com/google-research/android_world)]\n\n### 4.2 ⚖️ LLM-Based Evaluation \n\n#### 4.2.1 📌 LLM-as-a-Judge\n\n- (*Arxiv'24*) Towards Better Human-Agent Alignment: Assessing Task Utility in LLM-Powered Applications [[📝 Paper](https://arxiv.org/abs/2402.09015)] \n- (*Arxiv'24*) LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods [[📝 Paper](https://arxiv.org/abs/2412.05579)]\n- (*Arxiv'25*) LiveIdeaBench: Evaluating LLMs’ Divergent Thinking for Scientific Idea Generation with Minimal Context [[📝 Paper](https://arxiv.org/abs/2412.17596)] [[💻 Code](https://github.com/x66ccff/liveideabench)]\n- (*ACL'25*) Auto-Arena: Automating LLM Evaluations with Agent Peer Debate and Committee Voting [[📝 Paper](https://arxiv.org/abs/2405.20267)] [[💻 Code](https://github.com/DAMO-NLP-SG/Auto-Arena-LLMs)]\n- (*Arxiv'25*) MCTS-Judge: Test-Time Scaling in LLM-as-a-Judge for Code Correctness Evaluation [[📝 Paper](https://arxiv.org/abs/2502.12468)]\n\n#### 4.2.2 📌 Agent-as-a-Judge\n\n- (*Arxiv'24*) Agent-as-a-Judge: Evaluate Agents with Agents [[📝 Paper](https://arxiv.org/abs/2410.10934)] [[💻 Code](https://github.com/metauto-ai/agent-as-a-judge)]\n\n### 4.3 🛡 Safety, Alignment, and Robustness for Lifelong / Self-Evolving Agents\n\n- (*Arxiv'24*) AgentHarm: A Benchmark for Measuring Harmfulness of LLM Agents [[📝 Paper](https://arxiv.org/abs/2410.09024) ]\n- (*NeurIPS'24 – Datasets \u0026 Benchmarks*) RedCode: Risky Code Execution and Generation [[📝 Paper](https://arxiv.org/abs/2411.07781) ]\n- (*Arxiv'24*) MobileSafetyBench: Evaluating Safety of Autonomous Agents in Mobile Device Control [[📝 Paper](https://arxiv.org/abs/2410.17520)] [[💻 Code](https://mobilesafetybench.github.io/)]\n- (*Arxiv'23*) Do the Rewards Justify the Means? Measuring Trade-Offs Between Rewards and Ethical Behavior in the MACHIAVELLI Benchmark [[📝 Paper](https://arxiv.org/abs/2304.03279) ]\n- (*Arxiv'24*) R-Judge: Benchmarking Safety Risk Awareness for LLM Judges [[📝 Paper](https://arxiv.org/abs/2401.10019)] [[💻 Code](https://rjudgebench.github.io/)]\n- (*ACL'25*) SafeLawBench: Towards Safe Alignment of Large Language Models [[📝 Paper](https://arxiv.org/abs/2506.06636) ]\n- (*Arxiv'25*) Accuracy Paradox in Large Language Models: Regulating Hallucination Risks in Generative AI\n [[📝 Paper](https://www.arxiv.org/abs/2509.13345) ]\n- (*ICLR'25 Spotlight*) AutoDAN-Turbo: A Lifelong Agent for Strategy Self-Exploration to Jailbreak LLMs [[📝 Paper](https://arxiv.org/abs/2410.05295)] [[💻 Code](https://github.com/SaFoLab-WISC/AutoDAN-Turbo)]\n- (*ACL'25*) AGrail: A Lifelong Agent Guardrail with Effective and Adaptive Safety Detection [[📝 Paper](https://arxiv.org/abs/2502.11448)] [[💻 Code](https://github.com/SaFoLab-WISC/AGrail4Agent)]\n\n\n[![Star History Chart](https://api.star-history.com/svg?repos=EvoAgentX/Awesome-Self-Evolving-Agents\u0026Date\u0026type=Date)](https://www.star-history.com/#EvoAgentX/Awesome-Self-Evolving-Agents\u0026Date\u0026Date)\n\n\n## 📚 Citation\n\nIf you find this survey useful in your research and applications, please cite using this BibTeX:\n\n```\n@article{fang2025comprehensive,\n  title={A Comprehensive Survey of Self-Evolving AI Agents: A New Paradigm Bridging Foundation Models and Lifelong Agentic Systems},\n  author={Fang, Jinyuan and Peng, Yanwen and Zhang, Xi and Wang, Yingxu and Yi, Xinhao and Zhang, Guibin and Xu, Yi and Wu, Bin and Liu, Siwei and Li, Zihao and others},\n  journal={arXiv preprint arXiv:2508.07407},\n  year={2025}\n}\n```\n\n\n## ☕ Acknowledgement\n\nWe would like to thank Shuyu Guo for his valuable contributions to the early-stage exploration and literature review on agent optimisation.\n\n## ✉️ Contact Us\n\nIf you have any questions or suggestions, please feel free to contact us via:\n\nEmail: j.fang.2@research.gla.ac.uk and zaiqiao.meng@gmail.com\n","projects_url":"https://awesome.ecosyste.ms/api/v1/lists/evoagentx%2Fawesome-self-evolving-agents/projects"}