skill evolution
Imagine an AI agent following a set of instructions. Skill evolution lets it rewrite those instructions based on practice runs. The papers show this often backfires: the agent learns wrong lessons, deletes useful rules, or settles for a decent fix instead of a great one.
The shared lesson is that disciplined reflection beats raw trial and error. Some methods use a memory bank to distill experience across attempts, reducing noise. Others branch the search across multiple versions instead of updating a single line, or adaptively choose how to revise. Across benchmarks, these structured approaches consistently yielded more robust and transferable skills.
- Papers
- 42
- Released code
- 5
- First seen
- June 2026
- Latest
- Nov 2026
37 papers in the last two months, against 4 in the two before.
Start here
Three papers to read first, in the order they came out- 1Contrasting group executions pinpoints which agent skill caused success or failure
GRSE compares multiple executions of the same question to assign credit to the responsible agent and skill, outperforming prior methods.
Oct 2026 - 2AI agents can rewrite their improvement routines, outperforming human-written instructions
Skill∞ recursively improves both task skills and the meta-skill used to optimize them.
Oct 2026 - 3Language model and skill set co-evolve through response feedback
SkillDuet coordinates model parameter updates and skill retention using response feedback, achieving 98.57% on ALFWorld and 92.97% on WebShop.
Nov 2026
How fast it is moving
Often studied with
agent skillsLearned capabilities that enable AI agents to perform specialized tasks with improved effectiveness.26 togetherself-evolving agentAn agent that improves by reusing its own experience, memory, or self-generated feedback, with success depending on verification and design.19 togetherverificationChecking outputs for correctness before accepting them, often to catch errors or reward hacking.8 togetherWho is working on it
The papers
Most central to this idea first, not most recent.
- Top Universitycs.AI
AI agents can rewrite their improvement routines, outperforming human-written instructions
Skill∞ recursively improves both task skills and the meta-skill used to optimize them.
National University of Singapore, City University of Hong Kong · Oct 2026
- Chinese Techcs.CL
Contrasting group executions pinpoints which agent skill caused success or failure
GRSE compares multiple executions of the same question to assign credit to the responsible agent and skill, outperforming prior methods.
Mayo Clinic, University of Manchester · Oct 2026
- Chinese Techcs.AI
Language model and skill set co-evolve through response feedback
SkillDuet coordinates model parameter updates and skill retention using response feedback, achieving 98.57% on ALFWorld and 92.97% on WebShop.
Alibaba · Nov 2026
- Academiccs.CR
Security skills guide coding agents to only 11% fully correct vulnerability detection
Current security skills rarely make coding agents fully identify Linux kernel vulnerabilities, while failure-driven skill refinement substantially improves detection on unseen cases.
York University · Oct 2026
- Big Techcs.AI
Agent evolvers can learn their own search strategies across tasks
A learned campaign policy lets agent evolvers organize their own search and match or exceed hand-designed loops while transferring across tasks.
AWS AI Labs, Georgia Institute of Technology · Oct 2026
- Academiccs.AI
Agents Repair Executable Skill Packages Better With Code-Structure Guidance
SkillScriptBench shows AST-guided, scope-limited editing substantially improves reliable repair of executable agent skills while coordinating scripts and documentation.
The Hong Kong University of Science and Technology, Harbin Institute of Technology, Shenzhen, China · Oct 2026
- Big Techcs.AI
Mined agent skills need domain-specific evidence and format choices
Mined skill performance depends on task domain; workflows beat ontologies on one benchmark, outcome labels help on another, but no configuration wins everywhere.
Massachusetts Institute of Technology, Microsoft Corporation · Oct 2026
- Big Techcs.RO
Robots improve reusable manipulation skills through guided simulation practice
RPG improves robot task success from 28.6% to 95% through autonomous practice without updating model weights.
UC Berkeley, MIT · Oct 2026
- Chinese Techcs.AI
Regularization limits agent skill growth without sacrificing downstream performance
SkillEvoReg uses dropout, complexity regularization, and counterexample validation to prevent overfitting in agent skill evolution.
Huawei Noah’s Ark Lab · Sept 2026
- Big Techcs.AIcode
Adaptively choosing how to revise LLM skills outperforms fixed revision strategies
StraTune adaptively selects revision operators to improve self-evolving LLM skills, outperforming fixed operators across tasks.
The University of Texas at Austin, Amazon · Sept 2026
- Independentcs.CL
AI agents struggle to adapt learned workflows as tools change
An executable benchmark shows agents fail to adapt when tool ecosystems evolve; a Steward manager improves task success and skill maintenance.
Oct 2026
- Chinese Techcs.AI
Statistical gates help self-editing agents avoid lasting performance regressions
SAGE uses per-item paired comparisons and one-sided paired tests to accept skill edits only when wins are statistically reliable, reducing regressions in self-evolving LLM agents.
Peking University, University of Oklahoma · Sept 2026
- Chinese Techcs.AI
Contrasting multiple trajectories sharpens skill learning for LLMs
SkillCome improves skill evolution for language models by contrasting multiple successful and failed trajectories and using dual memory to derive reusable skills.
Fudan University, Ant Group · Sept 2026
- Chinese Techcs.AI
Branching search improves how language agents evolve reusable skill libraries
SkillVine uses branching exploration to evolve LLM agent skill libraries, outperforming linear evolution on nine of ten benchmark-model combinations.
The Chinese University of Hong Kong, Noah’s Ark Lab, Huawei Technologies · Sept 2026
- Academiccs.AI
Coding up definite operations improves AI agent consistency
Compiling well-specified skill operations into deterministic programs reduces execution uncertainty and improves AI agent task performance.
East China Normal University, China University of Geoscience · Oct 2026
- Big Techcs.RO
Humanoid tracker expands executable manipulation skills from sparse human demonstrations
InterMimicGen self-evolves motion imitation data and policies to scale humanoid loco-manipulation from sparse human demonstrations.
University of Illinois Urbana-Champaign, NVIDIA · Oct 2026
- Chinese Techcs.AI
Adaptive feedback helps reinforcement-learning agents target their changing weaknesses
ARISE adapts evaluation criteria and training priorities to an agent's evolving capabilities, improving long-horizon task performance.
ShanghaiTech University, Ant Group · Sept 2026
- Research Labcs.CV
Visual agents improve by jointly evolving reusable skills and practice data
V-Gym co-evolves procedural skills and multimodal practice data from agent execution trajectories for continual self-improvement.
State Key Laboratory of AI Safety, Institute of Computing TechnologyChinese Academy of Sciences · Sept 2026
- Chinese Techcs.AI
Adam-inspired updates make agent skill evolution steadier and more efficient
SkillAdam uses an Adam-inspired memory and adaptive budget to stabilize and accelerate skill evolution for LLM agents.
Renmin University of China, Tencent · Sept 2026
- Big Techcs.AI
Persistent Wikis Help AI Agents Build Better Transferable Skills
WikiSkill uses a persistent knowledge base to compile agent experience into reusable skills, outperforming existing methods.
Google Research, Virginia Tech · Aug 2026
- Academiccs.CR
Poisoned reference answers can implant persistent backdoors in AI agents
Poisoned reference answers can create persistent, trigger-activated backdoors in text-space skill optimization without attacker access to model or optimizer.
Yonsei University, Inha University · Nov 2026
- Chinese Techcs.AI
AI agents improve when training tasks adapt to their weaknesses
EvoMCP improves AI agents by co-evolving training tasks and demonstrations based on student feedback.
Shanghai Jiao Tong University, Ant Group · Nov 2026
- Chinese Techcs.AI
Interleaved teaching helps agents learn skills that transfer better
Interleaving expert teaching with student practice improves transferable skill learning for LLM agents.
Nanjing University, Alibaba Group · Oct 2026
- Big Techcs.AI
TeleTune learns agent skills from unfiltered user logs without goals or replay
TeleTune learns reusable textual skills from goal-free, non-replayable, interleaved telemetry logs by keeping only library edits that improve held-out action-prediction accuracy.
UNC Chapel Hill, University of Texas at Austin · Oct 2026
- Top Universitycs.AI
Language models can manage their own context more efficiently
Giving language models direct control over editable context improves long-horizon task performance while reducing inference compute compared with externally managed context.
University of Washington, Meta Superintelligence Labs · Sept 2026
- Chinese Techcs.LG
Refining rejected candidates makes AI system optimization more efficient
Mara Chain retains and refines rejected AI system candidates instead of discarding them, improving performance with fewer evaluations.
Ant Group, Beijing Intelligent Game and Decision Lab · Sept 2026
- Top Universitycs.RO
Visual action rehearsal helps language models control robot manipulation
A visual action workspace lets general-purpose VLMs rehearse, revise, and correct robot actions, achieving 75.6% average success on LIBERO-Pro.
HKUST(GZ), CUHK · Sept 2026
- Academiccs.AI
Subtask routing halves agent inference cost while improving performance
RSI-router achieves cheaper, better agent performance by routing subtasks to different LLMs and evolving skills.
Shanghai Artificial Intelligence Laboratory · Sept 2026
- Top Universitycs.AIcode
Bandit-guided skill evolution cuts LLM agent optimization costs
COBRA-Skills uses contextual bandits to optimize LLM agent skills, achieving top performance with 55-58% less cost than SkillOpt.
The Chinese University of Hong Kong, Shenzhen, Tianjin University · Sept 2026
- Top Universitycs.AIcode
EvoSteer Repairs Multi-Agent Workflows While They Are Still Running
EvoSteer outperforms baselines across twelve datasets with online self-evolving graph orchestration using AnchorTB and validated skill admission.
The Chinese University of Hong Kong, Shenzhen, Dalian University of Technology · Oct 2026
- Top Universitycs.RO
Robot agents improve by building and reusing executable motion tools
URAI improves robot manipulation by letting agents create reusable motion tools, execute them locally, and make model-level decisions between tool calls.
Hexafuture Inc., Peking University · Oct 2026
- Chinese Techcs.AI
MetaExp trains a transferable experience-learning capability in language model agents
MetaExp trains language models to learn and reuse experience across tasks using verifiable cross-task families.
University of Macau, Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences · Oct 2026
- Big Techcs.AI
A verifiable runtime for agents that extend themselves with safe, reversible tools.
A Lean 4 agent runtime uses reversibility to decide human oversight, letting agents extend their own tools and act autonomously.
Amazon · Oct 2026
- Chinese Techcs.AI
Dependency graphs pinpoint decisive errors in long agent execution traces
DeFA achieves the highest accuracy in attributing decisive errors and responsible agents in LLM agent trajectories.
Qwen DianJin Team, Alibaba Cloud Computing, Beihang University · Oct 2026
- Academic
Scoping Agent Memory Prevents Unrelated Tasks From Disrupting Each Other
Family-scoped retrieval prevents cross-family interference and supports repeated adaptation in persistent agent memory.
Independent, Northwestern University · Sept 2026
- Top Universitycs.CL
Self-programmed memory adapts retrieval depth to each agent query
MemCodex improves task success by 10.1% with 3.4x fewer tokens and 2.1x faster inference.
Université de Montréal, Mila – Quebec AI Institute · Oct 2026
- Big Techcs.LG
Competitive self-play discovers compact motor skills humans can directly control
Competitive self-play yields a small set of motor skills that humans can directly compose to solve new tasks without retraining.
Georgia Institute of Technology, Simon Fraser University · Oct 2026
- Big Techcs.AI
Learning options by focusing on relevant features instead of whole states
Identifying a small relevant subset of features per subgoal yields options that generalize broadly and accelerate exploration.
Amazon, Brown University · Sept 2026
- Industrycs.LGcode
Reusing agent trajectories cuts token use on repeated tool tasks
TRIAGE reduces token consumption in LLM agents by reusing historical execution trajectories through a three-level routing framework.
China Telecom Cloud · Sept 2026
- Chinese Techcs.CV
Video agents improve their own harnesses using unlabeled video
VideoEvoHarness evolves video agent harnesses from unlabeled videos without annotations or parameter updates, improving accuracy.
Institute of Automation, Chinese Academy of Sciences, University of Chinese Academy of Sciences · Oct 2026
- Chinese Techcs.AIcode
Evolutionary training harness co-evolves with LLM policies for RL
Co-evolving LLM policies with training-side diagnostic harnesses matches or exceeds human-engineered RL references across math, coding, and long-horizon software engineering.
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Tongyi Lab , Alibaba Group · June 2026
- Top Universitycs.AI
Agents improve long-horizon performance by evolving how they manage context
Evolving context management policies from long-horizon trajectories improves long-horizon agent performance more effectively than evolving skills or using fixed context strategies.
Fudan University, Shanghai Key Laboratory of Data Science · Sept 2026
Concepts are extracted from each paper and reused across the corpus, so this page grows on its own as the desk reads.