selfimproving-agent/Awesome-Self-Improving-Agents：基于基础模型的智能体系统自我提升策划列表

项目介绍

selfimproving-agent/Awesome-Self-Improving-Agents 是一个托管在 GitHub 上的开源项目，采用 MIT License 许可证，目前获得 621 个 Star、72 次 Fork。

仓库全名: selfimproving-agent/Awesome-Self-Improving-Agents

README 摘要

Awesome Self-Improving Modern Agentic Systems A curated and continuously evolving resource hub for self-improving agentic systems This repository brings together 📄 Papers · 📊 Benchmarks · ✍️ Blogs · 🎙️ Podcasts · 💬 Interviews · 🎥 Videos · 🧑‍🏫 Workshops & Courses Contributions are welcome — open a PR to share relevant resources and help grow the community! 🌱 alt="Read the paper" alt="Hugging Face Daily…

完整内容见 GitHub 原始 README 或本页。

常见问题

Awesome-Self-Improving-Agents 可以免费商用吗？

如何获取 Awesome-Self-Improving-Agents 的源码？

可以直接克隆仓库：git clone https://github.com/selfimproving-agent/Awesome-Self-Improving-Agents.git，默认分支为 main；安装与运行步骤见仓库 README。

Awesome-Self-Improving-Agents 最近还在维护吗？

仓库最近一次更新为 3 天前，当前没有未关闭的 Issue；Star 数为 621，可作为社区活跃度的参考。

Awesome Self-Improving Modern Agentic Systems

A curated and continuously evolving resource hub for self-improving agentic systems
_{This repository brings together
📄 Papers · 📊 Benchmarks · ✍️ Blogs · 🎙️ Podcasts ·
💬 Interviews · 🎥 Videos · 🧑‍🏫 Workshops & Courses}
_{Contributions are welcome — open a PR to share relevant resources and help grow the community! 🌱}

This repository accompanies Self-Improvements in Modern Agentic Systems: A Survey .

🌐 Contents

Definition & Scope
Survey Paper
Taxonomy
Key Literature
- 🟦 Foundation Model Improvement
- 🟩 Scaffolding Improvement
Evaluation & Benchmarking
Related Resources
Contact & Community
Contribute
Citation

🧩 Definition & Scope

This repository focuses on:

Foundation-Model-Based Agents: Autonomous systems that use foundation models as cognitive cores and operate through persistent scaffolds such as prompts, memory, tools, and control logic.
Self-Improvement Loops: Agent-level update loops in which signals or artifacts produced through generation, intrinsic evaluation, or environment interaction are consolidated into persistent components of the agent.
Foundation Model Improvement: Parameter-level updates driven by intrinsic generative demonstrations, intrinsic evaluative feedback, or extrinsic exploratory experience.
Scaffolding Improvement: Persistent updates to prompts, memory, tools, workflows, or full agent scaffolds.

📄 Survey Paper

Main figure of the survey
Overview of self-improvement mechanisms in foundation model–based agentic systems.

🧭 Taxonomy

If you are not yet familiar with agents, this simple illustration shows how an FM-based agent differs from a standalone foundation model (e.g., an LLM).

View the animated comparison

An illustration of the distinction between a standalone foundation model and an FM-Based agent.

To ground our taxonomy, we first introduce a formal abstraction of a foundation model–based agent, which serves as the basic unit for all self-improving agentic systems considered in this survey.

Formal abstraction of an FM-based agent
Formal abstraction of a foundation model–based agent, consisting of a foundation model and its scaffolding.

Based on this abstraction, we categorize self-improving agents along two orthogonal axes, depending on which component is improved during learning and adaptation.

🚀 Self-Improving Agents
- 🧠 Foundation Model Improvement
  - 📄 1.1 Intrinsic Generative Demonstrations
  - ⚖️ 1.2 Intrinsic Evaluative Feedback
  - 🌍 1.3 Extrinsic Exploratory Experience
    - 🤖 1.3.1 Interaction with Grounded Task Environments
    - 🔮 1.3.2 Interaction with Simulated Proxy Environments
- 🏗️ Scaffolding Improvement
  - ✍️ 2.1 Prompt Optimization
  - 💾 2.2 Memory
  - 🛠️ 2.3 Tool
  - ♾️ 2.4 Full Scaffolding

📚 Key Literature

🛣️ Evolution of Self-Improving Agents

Timeline of representative self-improving agent systems, organized by foundation model improvement and scaffolding improvement.

🔖 Papers List

🟦 Foundation Model Improvement

1.1 Intrinsic Generative Demonstrations

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Self-Instruct: Aligning Language Models with Self-Generated Instructions	ACL	paper	code
2023	Large Language Models Can Self-Improve	EMNLP	paper	N/A
2023	Orca: Progressive Learning from Complex Explanation Traces of GPT-4	arXiv	paper	code
2024	SELF: Self-Evolution with Language Feedback	arXiv	paper	N/A
2024	SELF-GUIDE: Better Task-Specific Instruction Following via Self-Synthetic Finetuning	COLM	paper	code
2025	Improving Model Alignment Through Collective Intelligence of Open-Source LLMS	ICML	paper	N/A
2025	Superficial Self-Improved Reasoners Benefit from Model Merging	EMNLP	paper	code
2025	Will Pre-Training Ever End? A First Step Toward Next-Generation Foundation MLLMs via Self-Improving Systematic Cognition	arXiv	paper	code
2025	TaskCraft: Automated Generation of Agentic Tasks	arXiv	paper	code
2025	Iterative Tool Usage Exploration for Multimodal Agents via Step-wise Preference Tuning	NeurIPS	paper	code
2025	Maximizing Confidence Alone Improves Reasoning	arXiv	paper	code
2025	DIVE: Diversified Iterative Self-Improvement	arXiv	paper	code
2025	Self-Adapting Language Models	NeurIPS	paper	code
2025	First SFT, Second RL, Third UPT: Continual Improving Multi-Modal LLM Reasoning via Unsupervised Post-Training	NeurIPS	paper	code
2025	LADDER: Self-Improving LLMs Through Recursive Problem Decomposition	arXiv	paper	N/A
2025	Self-Consistency Preference Optimization	ICML	paper	N/A
2025	Adapting While Learning: Grounding LLMs for Scientific Problems with Tool Usage Adaptation	ICML	paper	code
2026	Reinforcing General Reasoning Without Verifiers	ICLR	paper	code
2026	SAGE: Multi-Agent Self-Evolution for LLM Reasoning	arXiv	paper	N/A
2026	ANDES: Agent Native Data Evolving Synthesis Tool for Autonomous Instruction Alignment	arXiv	paper	code
2026	EvoGround: Self-Evolving Video Agents for Video Temporal Grounding	arXiv	paper	code
2026	SIMIT: Self-Improving Vision-Language Models via Imagination at Test-Time	arXiv	paper	code

1.2 Intrinsic Evaluative Feedback

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2022	Constitutional AI: Harmlessness from AI Feedback	arXiv	paper	code
2023	ReST meets ReAct: Self-Improvement for Multi-Step Reasoning LLM Agent	arXiv	paper	N/A
2025	STRIVE: Structured Reasoning for Self-Improvement in Claim Verification	MIR	paper	N/A
2025	Beyond Accuracy: The Role of Calibration in Self-Improving Large Language Models	arXiv	paper	N/A
2025	Self-Evolved Reward Learning for LLMs	ICLR	paper	code
2025	Sample, Predict, then Proceed: Self-Verification Sampling for Tool Use of LLMs	arXiv	paper	N/A
2025	RLSR: Reinforcement Learning from Self Reward	arXiv	paper	N/A
2025	Right Question is Already Half the Answer: Fully Unsupervised LLM Reasoning Incentivization	NeurIPS	paper	code
2025	TTRL: Test-Time Reinforcement Learning	NeurIPS	paper	code
2025	Can Large Reasoning Models Self-Train?	arXiv	paper	code
2025	Self Rewarding Self Improving	arXiv	paper	N/A
2025	Self-Evolving Curriculum for LLM Reasoning	arXiv	paper	code
2025	Reflect, Retry, Reward: Self-Improving LLMs via Reinforcement Learning	arXiv	paper	N/A
2025	Adaptive Self-improvement LLM Agentic System for ML Library Development	ICML	paper	code
2026	Learning to Reason without External Rewards	ICLR	paper	code
2026	Structured Reasoning for Large Language Models	arXiv	paper	N/A
2026	iReasoner: Trajectory-Aware Intrinsic Reasoning Supervision for Self-Evolving Large Multimodal Models	arXiv	paper	code
2026	STRIVE: Structured Reasoning for Self-improvement in Claim Verification	machine intelligence research	paper	N/A
2026	UniCorn: Towards Self-Improving Unified Multimodal Models through Intrinsic Evaluative Feedback	arXiv	paper	code
2026	Retrospective Progress-Aware Self-Refinement for LLM Agent Training	arXiv	paper	N/A
2026	EVE-Agent: Evidence-Verifiable Self-Evolving Agents	arXiv	paper	N/A

1.3 Extrinsic Exploratory Experience

1.3.1 Interaction with Grounded Task Environments

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	RoboCat: A Self-Improving Generalist Agent for Robotic Manipulation	TMLR	paper	code
2025	Tool-Star: Empowering LLM-Brained Multi-Tool Reasoner via Reinforcement Learning	arXiv	paper	code
2025	CodeARC: Benchmarking Reasoning Capabilities of LLM Agents for Inductive Program Synthesis	COLM	paper	code
2025	LLMs are Greedy Agents: Effects of RL Fine-tuning on Decision-Making Abilities	arXiv	paper	N/A
2025	Agent-RLVR: Training Software Engineering Agents via Guidance and Environment Rewards	arXiv	paper	N/A
2025	WebRL: Training LLM Web Agents via Self-Evolving Online Curriculum Reinforcement Learning	ICLR	paper	code
2025	Self-Improving Language Models for Evolutionary Program Synthesis: A Case Study on ARC-AGI	ICML	paper	code
2025	DeepResearcher: Scaling Deep Research via Reinforcement Learning in Real-world Environments	arXiv	paper	code
2025	Agentic Reasoning and Tool Integration for LLMs via Reinforcement Learning	arXiv	paper	N/A
2025	UI-Genie: A Self-Improving Approach for Iteratively Boosting MLLM-based Mobile GUI Agents	NeurIPS	paper	code
2025	RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learning	arXiv	paper	code
2025	SEAgent: Self-Evolving Computer Use Agent with Autonomous Learning from Experience	arXiv	paper	code
2026	WebGym: Scaling Training Environments for Visual Web Agents with Realistic Tasks	arXiv	paper	code
2026	Kevin: Multi-Turn RL for Generating CUDA Kernels	ICLR	paper	N/A
2026	Tool-R0: Self-Evolving LLM Agents for Tool-Learning from Zero Data	arXiv	paper	code
2026	Socratic-SWE: Self-Evolving Coding Agents via Trace-Derived Agent Skills	arXiv	paper	N/A
2026	Self-evolving LLM Agents with In-Distribution Optimization	ICML	paper	N/A
2026	Skill Self-Play: Pushing the Frontier of LLM Capability with Co-Evolving Skills	arXiv	paper	code
2026	DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training	arXiv	paper	code
2026	EnvHarness: Awakening Static Worlds for Agent Learning	arXiv	paper	code
2026	ReSAIL: Mitigating Collapse in Iterative Agent Self-Distillation	arXiv	paper	code
2026	Environmental Feedback Modeling Matters: Rethinking Feedback Treatment in Agentic Hindsight Self-Distillation	arXiv	paper	N/A

1.3.2 Interaction with Simulated Proxy Environments

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Language Models Meet World Models: Embodied Experiences Enhance Language Models	NeurIPS	paper	code
2024	Agent Planning with World Knowledge Model	NeurIPS	paper	code
2025	Web Agents with World Models: Learning and Leveraging Environment Dynamics in Web Navigation	ICLR	paper	code
2025	Understanding World or Predicting Future? A Comprehensive Survey of World Models	ACM CSUR	paper	code
2025	General agents contain world models	ICML	paper	N/A
2025	WebEvolver: Enhancing Web Agent Self-Improvement with Coevolving World Model	EMNLP	paper	code
2025	WebSynthesis: World-Model-Guided MCTS for Efficient WebUI-Trajectory Synthesis	arXiv	paper	code
2025	GAWM: Global-Aware World Model for Multi-Agent Reinforcement Learning	arXiv	paper	N/A
2025	WMPO: World Model-based Policy Optimization for Vision-Language-Action Models	arXiv	paper	code
2025	Internalizing World Models via Self-Play Finetuning for Agentic RL	arXiv	paper	N/A
2026	AlignUSER: Human-Aligned LLM Agents via World Models for Recommender System Evaluation	arXiv	paper	N/A
2026	WorldEvolver: Self-Evolving World Models for LLM Agent Planning	arXiv	paper	N/A
2025	NavMorph: A Self-Evolving World Model for Vision-and-Language Navigation in Continuous Environments	ICCV	paper	code
2025	EvoWorld: Evolving Panoramic World Generation with Explicit 3D Memory	arXiv	paper	N/A
2026	RISE: Self-Improving Robot Policy with Compositional World Model	RSS	paper	code

🟩 Scaffolding Improvement

2.1 Prompt Optimization

2.1.1 Scalar-Feedback Optimization

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2022	Large Language Models Are Human-Level Prompt Engineers	arXiv	paper	code
2024	Large Language Models as Optimizers	ICLR	paper	code
2024	Prompt Refinement with Image Pivot for Text-to-Image Generation	ACL	paper	code
2024	Learning from Contrastive Prompts: Automated Optimization and Adaptation	arXiv	paper	N/A
2024	PRompt Optimization in Multi-Step Tasks (PROMST): Integrating Human Feedback and Heuristic-based Sampling	EMNLP	paper	code
2025	The Prompt Alchemist: Automated LLM-Tailored Prompt Optimization for Test Case Generation	arXiv	paper	N/A
2025	DRO-InstructZero: Distributionally Robust Prompt Optimization for Large Language Models	arXiv	paper	N/A
2025	CoolPrompt: Automatic Prompt Optimization Framework for Large Language Models	FRUCT	paper	N/A
2026	SePO: Self-Evolving Prompt Agent for System Prompt Optimization	arXiv	paper	code
2026	SAGE: Stochastic Prompt Optimization via Agent-Guided Exploration	arXiv	paper	N/A

2.1.2 Qualitative-Feedback Refinement

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Self-Refine: Iterative Refinement with Self-Feedback	arXiv	paper	code
2023	Chain of Hindsight Aligns Language Models with Feedback	arXiv	paper	code
2023	Reflexion: Language Agents with Verbal Reinforcement Learning	NeurIPS	paper	code
2024	Self-Improving Customer Review Response Generation Based on LLMs	COLING	paper	N/A
2024	Prompt Optimization with Human Feedback	ICML Workshop	paper	code
2024	Optimizing Instructions and Demonstrations for Multi-Stage Language Model Programs	EMNLP	paper	code
2025	CriSPO: Multi-Aspect Critique-Suggestion-guided Automatic Prompt Optimization for Text Generation	AAAI	paper	code
2025	Boosting Private Domain Understanding of Efficient MLLMs: A Tuning-free, Adaptive, Universal Prompt Optimization Framework	arXiv	paper	N/A
2026	GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning	ICLR	paper	code
2026	From a Prompt to Repertoires: Evolving Functional REpertoires Enable LLM Continual Learning	arXiv	paper	N/A
2026	FORGE: Self-Evolving Agent Memory With No Weight Updates via Population Broadcast	CAIS	paper	N/A
2026	Recursive Harness Distillation across Agents for Robot Manipulation	arXiv	paper	N/A

2.1.3 Population-Based Evolution

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Promptbreeder: Self-Referential Self-Improvement Via Prompt Evolution	arXiv	paper	N/A
2024	EvoPrompt: Connecting LLMs with Evolutionary Algorithms Yields Powerful Prompt Optimizers	ICLR	paper	code
2024	WizardLM: Empowering large pre-trained language models to follow complex instructions	ICLR	paper	code
2025	Tournament of Prompts: Evolving LLM Instructions Through Structured Debates and Elo Ratings	KDD	paper	N/A
2025	DelvePO: Direction-Guided Self-Evolving Framework for Flexible Prompt Optimization	arXiv	paper	code
2025	How to Auto-optimize Prompts for Domain Tasks? Adaptive Prompting and Reasoning through Evolutionary Domain Knowledge Adaptation	NeurIPS	paper	code

2.1.4 Textual Gradient Optimization

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Automatic Prompt Optimization with “Gradient Descent” and Beam Search	EMNLP	paper	code
2024	How to Correctly do Semantic Backpropagation on Language-based Agentic Systems	arXiv	paper	code
2024	Trace is the Next AutoDiff: Generative Optimization with Rich Feedback, Execution Traces, and LLMs	NeurIPS	paper	code
2025	TextGrad: Automatic “Differentiation” via Text	Nature	paper	code
2025	metaTextGrad: Automatically optimizing language model optimizers	NeurIPS	paper	code
2025	MAPGD: Multi-Agent Prompt Gradient Descent for Collaborative Prompt Optimization	NeurIPS Workshop	paper	N/A
2025	Scaling Textual Gradients via Sampling-Based Momentum	ICML	paper	N/A
2026	Pick Your Textual Gradients	arXiv	paper	N/A
2026	SkillOpt: Executive Strategy for Self-Evolving Agent Skills	arXiv	paper	code
2026	VASO: Formally Verifiable Self-Evolving Skills for Physical AI Agents	arXiv	paper	N/A
2026	Learning to Evolve: A Self-Improving Framework for Multi-Agent Systems	ACL Findings	paper	N/A
2026	Learning to Learn-at-Test-Time: Language Agents with Learnable Adaptation Policies	arXiv	paper	code

2.2 Memory

2.2.1 Memory Object

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Learning to Reason and Memorize with Self-Notes	NeurIPS	paper	N/A
2024	ExpeL: LLM Agents Are Experiential Learners	AAAI	paper	code
2024	A Human-Inspired Reading Agent with Gist Memory of Very Long Contexts	ICML	paper	N/A
2024	CodeAgent: Enhancing Code Generation with Tool-Integrated Agent Systems for Real-World Repo-level Coding Challenges	ACL	paper	N/A
2024	MEMORYLLM: Towards Self-Updatable Large Language Models	ICML	paper	code
2025	Agent Workflow Memory	ICML	paper	code
2025	ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory	arXiv	paper	N/A
2025	Seeing, Listening, Remembering, and Reasoning: A Multimodal Agent with Long-Term Memory	arXiv	paper	code
2025	Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory	arXiv	paper	code
2025	PRIME: Planning and Retrieval-Integrated Memory for Enhanced Reasoning	AAAI	paper	N/A
2025	Contextual Memory Reweaving in Large Language Models Using Layered Latent State Reconstruction	arXiv	paper	code
2025	MemGen: Weaving Generative Latent Memory for Self-Evolving Agents	arXiv	paper	code
2025	M+: Extending MemoryLLM with Scalable Long-Term Memory	arXiv	paper	code
2026	Seeing, Listening, Remembering, and Reasoning: A Multimodal Agent with Long-Term Memory	ICLR	paper	code
2026	Thought-Retriever: Don’t Just Retrieve Raw Data, Retrieve Thoughts for Memory-Augmented Agentic Systems	arXiv	paper	N/A
2026	Better with Experience: Self-Evolving LLM Agents for Evidence-Grounded Health Community Notes	arXiv	paper	N/A
2026	Agon: An Autonomous Large-Scale Omnidisciplinary Research System Built on Prompt Economy	arXiv	paper	code
2026	Spotter: Let the Embodied Model Lead, and the VLM Reflect for It	arXiv	paper	code

2.2.2 Memory Structure

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2022	XMem: Long-Term Video Object Segmentation with an Atkinson-Shiffrin Memory Model	ECCV	paper	code
2023	Generative Agents: Interactive Simulacra of Human Behavior	UIST’23	paper	N/A
2024	MemoryBank: Enhancing Large Language Models with Long-Term Memory	arXiv	paper	code
2024	MovieChat: From Dense Token to Sparse Memory for Long Video Understanding	CVPR	paper	code
2024	Explore, Select, Derive, and Recall: Augmenting LLM with Human-like Memory for Mobile Task Automation	ACM MobiCom	paper	code
2025	SCM: Enhancing Large Language Model with Self-Controlled Memory Framework	DASFAA	paper	code
2025	Hierarchical Memory for High-Efficiency Long-Term Reasoning in LLM Agents	arXiv	paper	N/A
2025	SALM: A Multi-Agent Framework for Language Model-Driven Social Network Simulation	arXiv	paper	N/A
2025	Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory	arXiv	paper	code
2025	G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems	NeurIPS	paper	code
2025	Zep: A Temporal Knowledge Graph Architecture for Agent Memory	arXiv	paper	N/A
2025	SGMem: Sentence Graph Memory for Long-Term Conversational Agents	arXiv	paper	N/A
2025	CausalRAG: Integrating Causal Graphs into Retrieval-Augmented Generation	ACL	paper	code
2025	GraphVideoAgent: Enhancing Long-form Video Understanding with Entity Relation Graphs	MM’25	paper	N/A
2025	Decentralizing AI Memory: SHIMI, a Semantic Hierarchical Memory Index for Scalable Agent Reasoning	arXiv	paper	N/A
2025	From Knowledge to Noise: CTIM-Rover and the Pitfalls of Episodic Memory in Software Engineering Agents	ACL Workshop	paper	code
2025	In Prospect and Retrospect: Reflective Memory Management for Long-term Personalized Dialogue Agents	ACL	paper	N/A
2025	MrSteve: Instruction-Following Agents in Minecraft with What-Where-When Memory	ICLR	paper	code
2026	EvolveMem: Self-Evolving Memory Architecture via AutoResearch for LLM Agents	arXiv	paper	code
2026	SAGE: A Self-Evolving Agentic Graph-Memory Engine for Structure-Aware Associative Memory	arXiv	paper	N/A
2026	Prism: An Evolutionary Memory Substrate for Multi-Agent Open-Ended Discovery	arXiv	paper	N/A
2026	DecentMem: Self-Evolving Multi-Agent Systems via Decentralized Memory	arXiv	paper	N/A
2026	EXG: Self-Evolving Agents with Experience Graphs	arXiv	paper	N/A
2026	CLAG: Adaptive Memory Organization via Agent-Driven Clustering for Small Language Model Agents	ACL Findings	paper	code

2.2.3 Memory Processing

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Generative Agents: Interactive Simulacra of Human Behavior	UIST	paper	code
2024	WizardLM: Empowering large pre-trained language models to follow complex instructions	ICLR	paper	N/A
2025	SEDM: Scalable Self-Evolving Distributed Memory for Agents	ICLR	paper	code
2025	MemInsight: Autonomous Memory Augmentation for LLM Agents	arXiv	paper	N/A
2025	MemGen: Weaving Generative Latent Memory for Self-Evolving Agents	arXiv	paper	code
2025	A-MEM: Agentic Memory for LLM Agents	NeurIPS	paper	code
2025	G-Memory: Tracing Hierarchical Memory for Multi-Agent Systems	arXiv	paper	code
2025	Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory	arXiv	paper	code
2025	Agentic Retrieval-Augmented Generation: A Survey on Agentic RAG	arXiv	paper	code
2025	Memory OS of AI Agent	EMNLP	paper	code
2025	SCM: Enhancing Large Language Model with Self-Controlled Memory Framework	DASFAA	paper	code
2025	Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory	arXiv	paper	code
2025	MLC-Agent: Cognitive Model based on Memory-Learning Collaboration in LLM Empowered Agent Simulation Environment	arXiv	paper	N/A
2026	ForeDreamer: A Self-Evolving Dual-Agent Memory Architecture for Future Event Prediction	EMNLP Findings	paper	code
2026	Meta Context Engineering via Agentic Skill Evolution	ICML	paper	code
2026	SKILL-KD: Contrastive Skill Distillation for LLM Agents	arXiv	paper	code
2025	MemInsight: Autonomous Memory Augmentation for LLM Agents	arXiv	paper	N/A
2026	Agentic Context Engineering: Evolving Contexts for Self-Improving Language Models	ICLR	paper	code
2026	MemRL: Self-Evolving Agents via Runtime Reinforcement Learning on Episodic Memory	arXiv	paper	code
2026	TMEM: Scaling Self-Evolving Agents via Parametric Memory	arXiv	paper	N/A
2026	MemQ: Integrating Q-Learning into Self-Evolving Memory for LLM Agents	arXiv	paper	code
2026	Memory Beyond Recall: A Dual-Process Cognitive Memory System for Self-Evolving Agents	arXiv	paper	N/A
2026	AEL: Agent Evolving Learning for Open-Ended Environments	arXiv	paper	code
2026	Metis: Bridging Text and Code Memory for Self-Evolving Agents	arXiv	paper	N/A
2026	Mem^2Evolve: Towards Self-Evolving Agents via Co-Evolutionary Capability Expansion and Experience Distillation	arXiv	paper	code
2026	AnchorGUI: Asymmetric Memory for Dual-Scale Learning in GUI Navigation	ECCV	paper	N/A
2026	Learning from Repair Experience: Self-Evolving Agents for Automated Vulnerability Repair	ICSE	paper	code

2.3 Tool

2.3.1 Dynamic Tool Routing

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Voyager: An Open-Ended Embodied Agent with Large Language Models	arXiv	paper	code
2024	ToolNet: Connecting Large Language Models with Massive Tools via Tool Graph	arXiv	paper	N/A
2025	AgentOrchestra: Orchestrating Hierarchical Multi-Agent Intelligence with the Tool-Environment-Agent(TEA) Protocol	arXiv	paper	N/A
2025	MetaAgent: Toward Self-Evolving Agent via Tool Meta-Learning	arXiv	paper	code
2025	OrchDAG: Complex Tool Orchestration in Multi-Turn Interactions with Plan DAGs	NeurIPS Workshop	paper	N/A
2025	AutoTIR: Autonomous Tools Integrated Reasoning via Reinforcement Learning	arXiv	paper	code
2025	MCP-Flow: Facilitating LLM Agents to Master Real-World, Diverse and Scaling MCP Tools	arXiv	paper	code
2025	In-the-Flow Agentic System Optimization for Effective Planning and Tool Use	NeurIPS	paper	code
2025	MassTool: A Multi-Task Search-Based Tool Retrieval Framework for Large Language Models	arXiv	paper	code
2025	AgentOrchestra: Orchestrating Multi-Agent Intelligence with the Tool-Environment-Agent(TEA) Protocol	arXiv	paper	N/A
2025	Iterative Tool Usage Exploration for Multimodal Agents via Step-wise Preference Tuning	NeurIPS	paper	code
2025	Tool-Star: Empowering LLM-Brained Multi-Tool Reasoner via Reinforcement Learning	arXiv	paper	code
2025	MCP-Zero: Active Tool Discovery for Autonomous LLM Agents	arXiv	paper	code
2025	AskToAct: Enhancing LLMs Tool Use via Self-Correcting Clarification	arXiv	paper	N/A
2025	MemTool: Optimizing Short-Term Memory Management for Dynamic Tool Calling in LLM Agent Multi-Turn Conversations	arXiv	paper	N/A
2025	Tool-Planner: Task Planning with Clusters across Multiple Tools	ICLR	paper	code
2025	Tool-to-Agent Retrieval: Bridging Tools and Agents for Scalable LLM Multi-Agent Systems	arXiv	paper	N/A
2025	ToolGen: Unified Tool Retrieval and Calling via Generation	ICLR	paper	code
2026	ToolACE-R: Model-aware Iterative Training and Adaptive Refinement for Tool Learning	AAAI	paper	code
2026	DeepAgent: A General Reasoning Agent with Scalable Toolsets	WWW	paper	code
2026	DeepEyesV2: Toward Agentic Multimodal Model	ICLR	paper	code
2026	In-the-Flow Agentic System Optimization for Effective Planning and Tool Use	ICLR	paper	code
2026	GenericAgent: A Token-Efficient Self-Evolving LLM Agent via Contextual Information Density Maximization	arXiv	paper	code
2026	ANDES: Agent Native Data Evolving Synthesis Tool for Autonomous Instruction Alignment	arXiv	paper	code
2026	TabAgent: Distilling LLM Agent Components into Cheap Classifiers	arXiv	paper	code
2026	AutoTool: Efficient Tool Selection for Large Language Model Agents	AAAI	paper	code

2.3.2 Iterative Tool Refinement

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Voyager: An Open-Ended Embodied Agent with Large Language Models	arXiv	paper	code
2025	STELLA: Self-Evolving LLM Agent for Biomedical Research	arXiv	paper	code
2025	SkillWeaver: Web Agents can Self-Improve by Discovering and Honing Skills	arXiv	paper	code
2025	PyVision: Agentic Vision with Dynamic Tooling	arXiv	paper	code
2025	From Exploration to Mastery: Enabling LLMs to Master Tools via Self-Driven Interactions	ICLR	paper	code
2025	LLMLOOP: Improving LLM-Generated Code and Tests Through Automated Iterative Feedback Loops	ICSME	paper	code
2025	Helping LLMs Improve Code Generation Using Feedback from Testing and Static Analysis	arXiv	paper	N/A
2026	RewardHarness: Self-Evolving Agentic Post-Training	arXiv	paper	code
2026	MUSE-Autoskill: Self-Evolving Agents via Skill Creation, Memory, Management, and Evaluation	arXiv	paper	N/A
2026	CODESKILL: Learning Self-Evolving Skills for Coding Agents	arXiv	paper	N/A
2026	PFAgent: A Tractable and Self-Evolving Power-Flow Agent for Interactive Grid Analysis	arXiv	paper	N/A

2.3.3 Autonomous Tool Creation

View

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Voyager: An Open-Ended Embodied Agent with Large Language Models	arXiv	paper	code
2024	Large Language Models as Tool Makers	arXiv	paper	code
2024	OS-Copilot: Towards Generalist Computer Agents with Self-Improvement	ICLR	paper	code
2025	Alita: Generalist Agent Enabling Scalable Agentic Reasoning with Minimal Predefinition and Maximal Self-Evolution	arXiv	paper	code
2025	Alita-G: Self-Evolving Generative Agent for Agent Generation	arXiv	paper	N/A
2025	LLM Agents Making Agent Tools	ACL	paper	code
2025	Advanced Tool Learning and Selection System (ATLASS): A Closed-Loop Framework Using LLM	SOSE	paper	N/A
2025	Code2MCP: Transforming Code Repositories into MCP Services	arXiv	paper	code
2025	STELLA: Self-Evolving LLM Agent for Biomedical Research	arXiv	paper	code
2025	PyVision: Agentic Vision with Dynamic Tooling	arXiv	paper	code
2025	AgentOrchestra: Orchestrating Multi-Agent Intelligence with the Tool-Environment-Agent(TEA) Protocol	arXiv	paper	N/A
2025	Enhancing Open-Domain Task-Solving Capability of LLMs via Autonomous Tool Integration from GitHub	ACL	paper	code
2026	OpenSkill: Open-World Self-Evolution for LLM Agents	arXiv	paper	code
2026	EvoDS: Self-Evolving Autonomous Data Science Agent with Skill Learning and Context Management	arXiv	paper	code
2026	Autonomous Evolution of EDA Tools: Multi-Agent Self-Evolved ABC	DAC	paper	N/A
2026	CoEvoSkills: Self-Evolving Agent Skills via Co-Evolutionary Verification	arXiv	paper	code

2.4 Full Scaffolding

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2024	Language Agents as Optimizable Graphs	ICML	paper	code
2024	Self-Taught Optimizer (STOP): Recursively Self-Improving Code Generation	COLM	paper	code
2024	Automated Design of Agentic Systems	NeurIPS	paper	code
2024	Symbolic Learning Enables Self-Evolving Agents	arXiv	paper	code
2025	Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents	arXiv	paper	code
2025	Huxley-Gödel Machine: Human-Level Coding Agent Development by an Approximation of the Optimal Self-Improving Machine	arXiv	paper	code
2025	Gödel Agent: A Self-Referential Agent Framework for Recursive Self-Improvement	ACL	paper	code
2025	AlphaEvolve: A coding agent for scientific and algorithmic discovery	arXiv	paper	N/A
2025	ShinkaEvolve: Towards Open-Ended And Sample-Efficient Program Evolution	arXiv	paper	code
2025	Live-SWE-agent: Can Software Engineering Agents Self-Evolve on the Fly?	arXiv	paper	code
2026	AgentDevel: Reframing Self-Evolving LLM Agents as Release Engineering	arXiv	paper	N/A
2026	JudgeFlow: Agentic Workflow Optimization via Block Judge	arXiv	paper	N/A
2026	RoboPhD: Self-Improving Text-to-SQL Through Autonomous Agent Evolution	arXiv	paper	code
2026	Group-Evolving Agents: Open-Ended Self-Improvement via Experience Sharing	arXiv	paper	N/A
2026	Hyperagents	arXiv	paper	code
2026	CORAL: Towards Autonomous Multi-Agent Evolution for Open-Ended Discovery	COLM	paper	code
2026	Adaptive Auto-Harness: Sustained Self-Improvement for Agentic System Deployment on Open-Ended Task Streams	arXiv	paper	code
2026	MOSS: Self-Evolution through Source-Level Rewriting in Autonomous Agent Systems	arXiv	paper	code
2026	Recursive Self-Evolving Agents via Held-Out Selection	arXiv	paper	N/A
2026	Continual Harness: Online Adaptation for Self-Improving Foundation Agents	arXiv	paper	code
2026	Argus: A General-Purpose Agentic Reasoning Runtime for Long-Horizon Tasks	arXiv	paper	code
2026	The Red Queen Gödel Machine: Co-Evolving Agents and Their Evaluators	arXiv	paper	N/A
2026	Harness-R1: Learning to Edit Executable Runtime Harnesses from Agent Failure Trajectories	arXiv	paper	code
2026	LLM-as-Code: Agentic Programming for Agent Harness	KDD AgenticSE Workshop	paper	code
2026	Ouroboros: A Self-Developing Frontier Coding Agent with Reviewed Core Evolution	arXiv	paper	code
2026	Harness Handbook: Making Evolving Agent Harnesses Readable, Navigable, and Editable	arXiv	paper	code
2026	Meta^n: Recursive Self-Improvement through Emergent Depth	arXiv	paper	code
2026	Prime Agent: A Self-Improving RLM Harness	arXiv	paper	code
2026	HarnessDev: Can LLMs Create and Evolve Their Own Agent Harness?	arXiv	paper	N/A
2026	EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents	arXiv	paper	code
2026	MetaRSI / RSI2: A Meta-Recursive Self-Improving System for Recursive Self-Improving Systems Themselves	arXiv	paper	code
2026	Harness-Zero: Harness Distillation via Agent-as-Harness	arXiv	paper	code
2026	SelfSearch: Reward-Free Search for Self-Improving Agents	arXiv	paper	N/A
2026	AutoRef: Harness Optimization for Agentic Multi-Reference Image Generation	arXiv	paper	code
2026	ReCreate: Reasoning and Creating Domain Agents Driven by Experience	ACL	paper	code
2026	Evolve as a Team: Collaborative Self-Evolution for LLM-based Multi-Agent Systems	arXiv	paper	code

📊 Evaluation & Benchmarking

A representative paper–benchmark matrix for self-improving agents, illustrating which benchmarks are used across different methods. See the Evaluation section of our survey for further details.

The literature is organized along two complementary perspectives:

Measuring Improvement studies how agent progress is assessed, either through explicit metrics or learned/agentic judges.
Benchmarking Improvement groups benchmarks by the improvement mechanism they probe or by the application domain they cover.

Note 注意

A benchmark may appear in more than one subsection because the taxonomy is multi-perspective. For example, a benchmark can provide metric-based measurement while also targeting a specific domain or improvement mechanism.

📏 Measuring Improvement

Metric-Based Measurement

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2023	Mind2Web: Towards a Generalist Agent for the Web	NeurIPS	paper	code
2023	ManiSkill2: A Unified Benchmark for Generalizable Manipulation Skills	ICLR	paper	code
2024	CORE-Bench: Fostering the Credibility of Published Research Through a Computational Reproducibility Agent Benchmark	TMLR	paper	code
2024	SWE-Bench+: Enhanced Coding Benchmark for LLMs	arXiv	paper	N/A
2024	WebLINX: Real-World Website Navigation with Multi-Turn Dialogue	ICML	paper	code
2024	GAIA: A Benchmark for General AI Assistants	ICLR	paper	N/A
2024	MINT: Evaluating LLMs in Multi-Turn Interaction with Tools and Language Feedback	ICLR	paper	code
2024	WorkArena: How Capable Are Web Agents at Solving Common Knowledge Work Tasks?	ICML	paper	code
2025	AgentGym: Evolving Large Language Model-Based Agents across Diverse Environments	ACL	paper	code
2025	GitTaskBench: A Benchmark for Code Agents Solving Real-World Tasks Through Code Repository Leveraging	arXiv	paper	code
2025	SafeAgentBench: A Benchmark for Safe Task Planning of Embodied LLM Agents	arXiv	paper	code
2026	DrunkAgent: Stealthy Memory Corruption in LLM-Powered Recommender Agents	WWW	paper	N/A
2026	ST-WebAgentBench: A Benchmark for Evaluating Safety and Trustworthiness in Web Agents	ICLR	paper	code
2026	ClawBench: A Benchmark for Evaluating AI Agents on Real-World Online Tasks	arXiv	paper	code

Judge-Based Measurement

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2025	Agent-as-a-Judge: Evaluate Agents with Agents	ICML	paper	code
2025	Evaluation Agent: Efficient and Promptable Evaluation Framework for Visual Generative Models	ACL	paper	code
2025	EvalAgent: Discovering Implicit Evaluation Criteria from the Web	COLM	paper	code
2025	Learning to Align Multi-Faceted Evaluation: A Unified and Robust Framework (ARJudge)	ACL Findings	paper	N/A
2025	VerifiAgent: A Unified Verification Agent in Language Model Reasoning	EMNLP Findings	paper	code

🧪 Benchmarking Improvement

Mechanism Benchmarks

Foundation-Model-Level

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2024	SWE-Bench+: Enhanced Coding Benchmark for LLMs	arXiv	paper	N/A
2024	Identifying the Risks of LM Agents with an LM-Emulated Sandbox	ICLR	paper	code
2025	GitTaskBench: A Benchmark for Code Agents Solving Real-World Tasks Through Code Repository Leveraging	arXiv	paper	code

Scaffold-Level

📅 Year	📝 Title	🏛️ Venue	📄 Paper	💻 Code
2024	MINT: Evaluating LLMs in Multi-Turn Interaction with Tools and Language Feedback	ICLR	paper	code
2024	TaskBench: Benchmarking Large Language Models for Task Automation	NeurIPS	paper	code
2024	MetaTool Benchmark for Large Language Models: Deciding Whether to Use Tools and Which to Use	ICLR	paper	code
2025	The Berkeley Function Calling Leaderboard (BFCL): From Tool Use to Agentic Evaluation of Large Language Models	ICML	paper	code
2026	DrunkAgent: Stealthy Memory Corruption in LLM-Powered Recommender Agents	WWW	paper	N/A
2026	RSI-Bench: Multi-Axis Benchmark for Recursive Self-Improvement	GitHub	N/A	code
2026	PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents	arXiv	[paper]	code

README 内容过长，已截取前 64 KB。完整内容请前往 GitHub 仓库查看。

Awesome-Self-Improving-Agents

项目介绍

README 摘要

常见问题

关于

简要分析

语言占比

贡献者

Awesome Self-Improving Modern Agentic Systems

🌐 Contents

🧩 Definition & Scope

📄 Survey Paper

🧭 Taxonomy

📚 Key Literature

🛣️ Evolution of Self-Improving Agents

🔖 Papers List

🟦 Foundation Model Improvement

🟩 Scaffolding Improvement

📊 Evaluation & Benchmarking

📏 Measuring Improvement

🧪 Benchmarking Improvement

Foundation-Model-Level

Scaffold-Level

评论

Awesome-Self-Improving-Agents

项目介绍

README 摘要

常见问题

相关仓库

关于

简要分析

语言占比

贡献者

Awesome Self-Improving Modern Agentic Systems

🌐 Contents

🧩 Definition & Scope

📄 Survey Paper

🧭 Taxonomy

📚 Key Literature

🛣️ Evolution of Self-Improving Agents

🔖 Papers List

🟦 Foundation Model Improvement

🟩 Scaffolding Improvement

📊 Evaluation & Benchmarking

📏 Measuring Improvement

🧪 Benchmarking Improvement

Foundation-Model-Level

Scaffold-Level

评论