Computation and Language · arXiv:2611.25429 v1 · submitted 19 Nov 2026 · 34 references
Evolve Where It Matters: Coverage-Guided Self-Evolution of Agent Skills
The paper says
“Agent skills package procedural knowledge as natural-language instructions, and a skill helps only when its instructions are correct and properly followed by the agent.”
more of the abstract
Self-evolving agents revise their own skills from the runs they make, but existing methods judge each run as a whole, by its outcome or by where it went wrong, and cannot tell whether a failure comes from a wrong instruction or from one the agent did not follow. Instruction-level coverage was introduced to test skills and labels each instruction of a run as follow, not-follow, or not covered, which separates these causes. We leverage it for skill self-evolution with SkillPlumber, a framework in which an agent evolves its own skills where coverage shows they fall short. A persistent root cause analysis agent turns each label into a revision question and checks each revision against the next run, while a maintainer distills investigations across tasks into a shared strategy library. Across three benchmarks and three executors, SkillPlumber solves 64.7% to 70.5% of the tasks, 25.6 to 30.4 percentage points higher than the initial skill and 8.7 to 14.0 percentage points higher than the strongest baseline, and removing coverage reduces recovered failures by 20.8% on SkillsBench with GPT-5.5.
from the abstract
Pith says
Machine review · deepseek-v4-flash · 20 Nov 2026 · read the full text (79,321 characters) · not checked by a person
It can be wrong. Challenge this review or reply to it.
Holds up. SkillPlumber turns instruction-level coverage—a label for how an agent uses each instruction—into a revision signal and shows, across three benchmarks and three executors, that it recovers more failures than existing skill self-evolution methods. This line is wrong
4 minor
How often objections from reviews like this one are right: not yet measured
The paper, first character to last
The reviewer was given the whole text, 79,321 characters. Not checked: figures, code, what the 34 references say.
4 minor objections
The paper says, at Abstract / Section 1
Pith says: minor objection 1 of 4
The labels "follow", "not-follow", and "not covered" are used throughout. Consider clarifying early on that these are past-tense descriptions (e.g., "followed", "not followed") for readability, especially for readers unfamiliar with the prior work on Skill Coverage.
The paper says, at Section 3, Algorithm 1
Pith says: minor objection 2 of 4
The algorithm uses 'Pass(y_t)' as the loop condition. It would be helpful to state explicitly that this checks the verifier output, since earlier the text defines y_t as verifier feedback. The current text in §4.1 says 'Pass(y_t) holds when this feedback reports that V(x,τ_t) is a pass', which is clear but appears after the algorithm.
show 2 more
The paper says, at Section 5.1, Table 1
Pith says: minor objection 3 of 4
The table reports 'Succ.' as raw fractions. For clarity, consider adding a column with the percentage or the percentage-point gain relative to 'No skill' in the table body rather than only in the Δpp column.
The paper says, at Section 5.1, Figure 2 (caption)
Pith says: minor objection 4 of 4
The caption of the right-side subfigure uses 'Table~\ref{tab:revisions}' which is fine, but the figure itself is not visible in the supplied text. The caption should state the full meaning of the x-axis (revision round number) for completeness.
What would settle it
A human evaluation of the coverage judge on a sample of SkillsBench tasks, showing that the judge frequently mislabels instructions (e.g., labeling a followed instruction as not-follow), would contradict the method's assumption that coverage provides a reliable signal. Alternatively, an experiment where coverage labels are replaced with random labels of the same distribution and the method still matches SkillPlumber's recovery would suggest the gain is not due to the label content.
from Pith's summary of the paper
How the review went
5 passes by deepseek-v4-flash · 20 Nov 2026
- Reader holds up
- Skeptic concern not confirmed: listed as an open question
- Referee 4 minor
- Check against the paper kept 0 of 1 objection
- Editor wrote the verdict above
Not run on this paper: the simulated authors' defence.
Pith's passes raised 1 objection. A check against the paper's text kept 0 and left 1 as a question.
The review in full
The review in brief, machine-written
This paper does something straightforward that turns out to be effective: it takes instruction-level coverage (already used for testing skills) and makes it the guide for skill self-evolution. Instead of judging a whole run as pass or fail, SkillPlumber labels each instruction as follow, not-follow, or not covered, then asks a persistent root-cause-analysis agent to revise where the labels point. A cross-task maintainer shares strategies across tasks. The result: on 207 tasks across SkillsBench, SkillLearnBench, and SWE-Skills-Bench-Hard, SkillPlumber solves the most tasks with all three executors (GPT-5.5, Opus 4.8, DeepSeek-V4-Pro). The gains are consistent; removing coverage costs 20.8% of recovered failures. The experiments are well controlled: same initial skills, same protocol, ablation with a trajectory-analysis control, replay tests, statistical tests. The paper is clearly written and honest about limitations (judge accuracy, token cost, benchmark version differences). The central claim—that coverage-guided revision outperforms outcome-only revision—holds up well.
The main soft spot is that the coverage labels come from an LLM judge, which can mislabel. The authors acknowledge this and cite validation from prior work, but mislabeling could steer revision in wrong directions. The effect is bounded by the persistent RCA agent that can challenge labels, but the paper does not measure how often mislabeling occurs or whether it harms specific cases. The maintainer contributes only one extra recovery, so the strategy library is not a major driver. The setting (same task for revision and evaluation) follows prior work and is appropriate for reused skills, but it means the results do not test generalization to unseen tasks.
This paper is for researchers working on LLM agent self-improvement, skill libraries, or test-time adaptation. It deserves a serious referee. The evidence is solid, the method is practical, and the framing of coverage as a revision signal is a genuine step over outcome-only approaches. Recommend peer review.
The stress test's concern
Not confirmed. Pith could not check this concern against the text of the paper it was given; it is listed as an open question, not a finding.
The referee's report
If the results hold, this work significantly advances skill self-evolution by introducing a principled signal (instruction-level coverage) that separates whether a failure is due to a wrong instruction or a non-followed one. The experimental design is thorough: three executors, controlled baselines sharing the same initial skill, replay tests, cross-model transfer, and statistical tests (bootstrap, McNemar). The paper identifies a clear limitation of prior methods (whole-run feedback) and addresses it with a concept borrowed from software testing. The findings are likely to influence future work on agent debugging, skill libraries, and self-improving systems.
Pith's reading of the paper (a machine's summary, not the authors')
Agent skills are natural-language instructions that an agent follows during a task. A skill helps only when its instructions are correct and properly followed. Existing self-evolution methods judge an entire run as pass or fail, but cannot tell whether a failure came from a wrong instruction or from the agent ignoring a correct one. SkillPlumber leverages instruction-level coverage, which labels each instruction as follow, not-follow, or not covered, to separate these causes. A persistent root cause analysis agent turns each label into a revision question and checks revisions against subsequent runs; a maintainer distills strategies across tasks. The authors claim that SkillPlumber solves 64.7% to 70.5% of tasks across three benchmarks and three executors, 25.6 to 30.4 percentage points above the initial skill and 8.7 to 14.0 percentage points above the strongest baseline, and that removing coverage reduces recovered failures by 20.8% on SkillsBench with GPT-5.5.
What carries the argument. Instruction-level coverage (Skill Coverage) labels each instruction of a skill as follow (the condition occurred and the agent acted as required), not-follow (the condition occurred but the agent did not comply), or not covered (the condition never arose). SkillPlumber turns these labels into revision questions: for a not-follow instruction, ask why it was not followed; for a followed instruction, ask whether it is itself wrong; for a not covered instruction, assess whether it is relevant. A persistent root cause analysis agent reviews its own previous hypotheses and predictions against the next run, and a maintainer accumulates investigative strategies across tasks.
Core claim. The paper claims that instruction-level coverage, which labels each skill instruction as follow, not-follow, or not covered from an execution trace, can be used as a signal for skill self-evolution. By directing revision toward the instructions the agent did not follow, rather than judging the run as a whole, SkillPlumber recovers more failed tasks than methods that rely on outcome-only or trajectory-level diagnosis. The paper further claims that the coverage signal, not the agentic revision budget or the strategy library, drives the gain: removing coverage reduces recoveries, while removing the maintainer has little effect.
The premise it rests on. The coverage labels (follow, not-follow, not covered) come from an LLM judge that may mislabel an instruction, and if the judge consistently mislabels, the root cause analysis agent will pursue wrong revision directions, potentially harming performance or wasting the revision budget.
If the paper is right (Pith's inferences, not the authors')
If SkillPlumber’s coverage-guided revision is correct, then self-evolving agents can distinguish between instruction errors and execution errors, leading to more reliable skill improvement without human intervention.
The method provides a way to use execution traces from deployment (where no verifier exists) to improve skills, since coverage requires only the trace, not the outcome.
The persistent RCA agent can retract its own earlier revisions when they do not work, which may reduce the number of wasted revision rounds compared to methods that always spend the full budget.
Combining instruction-level coverage with code coverage of attached scripts could allow revision of the entire skill bundle, not just natural-language instructions.
The approach treats skill revision as analogous to program debugging with code coverage; this analogy may inspire further work on test adequacy metrics for agent behaviour, such as branch coverage or mutation testing for skill instructions.
Because the coverage judge is the same model as the executor, mislabelling could be systematic; future work could explore using a separate, specialised judge or incorporating human-in-the-loop verification for critical instructions.
The cross-model transfer experiment suggests that coverage-driven revisions capture some executor-specific behaviour; reusing skills across different models may require re-running the revision loop for each target model, not just transferring final skills.
Check this review
What was read
The passes were given all 79,321 characters of the paper's LaTeX source.
Fingerprint of that text (SHA-256): b84a5e7cda19dc29a72173c19fff373c9a12fd36d70209e9effc2b661590b22c
Pith does not keep that text. Rebuilt from the paper's source or PDF by the same code, it has this fingerprint if it is the same text. From a copy of Pith's code:
PYTHONPATH=. python scripts/review_input_check.py 2611.25429
The passes, in the order they ran
- Reader deepseek-v4-flash · medium effort · $0.0017 · instructions of the Reader, fingerprint f9ead7300382…
- Skeptic deepseek-v4-flash · medium effort · $0.0015 · instructions of the Skeptic, fingerprint 8f88c0f61488…
- Referee deepseek-v4-flash · medium effort · $0.0015 · instructions of the Referee, fingerprint a852d99aea0d…
- Check against the paper deepseek-v4-flash · medium effort · $0.0016 · instructions of the Check against the paper, fingerprint 6c7a6883d551…
- Editor deepseek-v4-flash · medium effort · $0.0018 · instructions of the Editor, fingerprint e7211f6a3d27…
- Circularity check deepseek-v4-flash · medium effort · $0.0013 · instructions of the Circularity check, fingerprint e9946e39d67b…
- Summary for the page deepseek-v4-flash · medium effort · temperature 0.2 · $0.0015 · instructions of the Summary for the page, fingerprint e4f56bf568e7…
The passes cost $0.011 in model calls together.
Version stamped on this review: pith-pipeline@v1.4.1-daily-deepseek-or-medium+deepseek/deepseek-v4-flash+p.d60f54226f00
Each link opens a pass's instructions as plain text, the exact bytes its fingerprint was taken over. Every stored set of instructions.
curl -s https://pith.science/methodology/prompts/f9ead730038216984cfb56d7cb34332bdf691828296a1de4245f3d6667d4bf56 | sha256sum
prints the fingerprint of the reader's instructions. The same works for every pass.
The receipt
The review's fingerprint was taken on 20 Nov 2026. The review you are reading still matches it.
Fingerprint of the review (SHA-256): 3236d26d10facc25b23cb915eb88c4d553fc682688b85c889e8aa9f8b27cf4e3
curl -s 'https://pith.science/api/v1/verdicts/2611.25429?form=canonical' | sha256sum
prints the fingerprint of the review as it stands now. The rule, in four steps.
A matching fingerprint shows the review has not changed since that date. It does not show that the review is right.
How often reviews like this one are right
Reviews like this one: written by deepseek-v4-flash, from the paper's LaTeX source.
How often objections from reviews like this one are right: not yet measured
Reviews by people
This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.
The paper
Abstract
Agent skills package procedural knowledge as natural-language instructions, and a skill helps only when its instructions are correct and properly followed by the agent. Self-evolving agents revise their own skills from the runs they make, but existing methods judge each run as a whole, by its outcome or by where it went wrong, and cannot tell whether a failure comes from a wrong instruction or from one the agent did not follow. Instruction-level coverage was introduced to test skills and labels each instruction of a run as follow, not-follow, or not covered, which separates these causes. We leverage it for skill self-evolution with SkillPlumber, a framework in which an agent evolves its own skills where coverage shows they fall short. A persistent root cause analysis agent turns each label into a revision question and checks each revision against the next run, while a maintainer distills investigations across tasks into a shared strategy library. Across three benchmarks and three executors, SkillPlumber solves 64.7% to 70.5% of the tasks, 25.6 to 30.4 percentage points higher than the initial skill and 8.7 to 14.0 percentage points higher than the strongest baseline, and removing coverage reduces recovered failures by 20.8% on SkillsBench with GPT-5.5.
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. No sentence in this record was written by a model.
abstract compiled · 2026-11-20
The authors' main claim, as the classifier picked it out 1.00
“Across three benchmarks and three executors, SkillPlumber solves 64.7% to 70.5% of the tasks, 25.6 to 30.4 percentage points higher than the initial skill and 8.7 to 14.0 percentage points higher than the strongest baseline, and removing coverage reduces recovered failures by 20.8% on SkillsBench with GPT-5.5.”
How the main claim is stated: asserted 1.00
What the abstract itself states, as probabilities: claims first 0.08 · states limitation 0.23 · human participants 0.04 · controlled comparison 0.95 · randomized assignment 0.02 · states quantitative result 0.99 · compares with prior results 0.98 · mentions code or data availability 0.04
The abstract, sentence by sentence
This table labels the abstract only (arXiv listing abstract v1). The number is the probability the classifier gave the label.
| Sentence | Role | The authors' own work | States the main conclusion | Strength |
|---|---|---|---|---|
| Agent skills package procedural knowledge as natural-language instructions, and a skill helps only when its instructions are correct and properly followed by the agent. | background 1.00 | 0.07 | 0.03 | asserted 1.00 |
| Self-evolving agents revise their own skills from the runs they make, but existing methods judge each run as a whole, by its outcome or by where it went wrong, and cannot tell whether a failure comes from a wrong instruction or from one the agent did not follow. | gap 0.95 | 0.07 | 0.03 | asserted 1.00 |
| Instruction-level coverage was introduced to test skills and labels each instruction of a run as follow, not-follow, or not covered, which separates these causes. | background 0.89 | 0.09 | 0.04 | asserted 1.00 |
| We leverage it for skill self-evolution with SkillPlumber, a framework in which an agent evolves its own skills where coverage shows they fall short. | contribution 0.85 | 0.90 | 0.11 | asserted 0.99 |
| A persistent root cause analysis agent turns each label into a revision question and checks each revision against the next run, while a maintainer distills investigations across tasks into a shared strategy library. | method 0.98 | 0.83 | 0.05 | asserted 0.99 |
| Across three benchmarks and three executors, SkillPlumber solves 64.7% to 70.5% of the tasks, 25.6 to 30.4 percentage points higher than the initial skill and 8.7 to 14.0 percentage points higher than the strongest baseline, and removing coverage reduces recovered failures by 20.8% on SkillsBench with GPT-5.5. | result 1.00 | 0.96 | 0.74 | asserted 1.00 |
What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.97
What was examined, and what was not
Sources read
- arXiv listing abstract v1
Not examined
- full text
- figures and tables
- references and citing sentences
- relations to other papers
Levels run
- P abstract
Not run
- S sections
- T body sentences
- C citations
- R cross-paper relations
How the labels were made
Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-11-20: 1 request, 5,195 input tokens.
Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. This record as JSON
Figures (5)
Figures are the authors', reproduced from the paper.
References (34)
- [1] Lakshya A. Agrawal, Shangyin Tan, Dilara Soylu, Noah Ziems, Rishi Khare, Krista Opsahl-Ong, Arnav Singhvi, Herumb Shandilya, Michael J Ryan, Meng Jiang, Christopher Potts, Koushik Sen, A. Dimakis, Ion Stoica, Dan Klein, Matei A. Zaharia, and O. Khattab. 2025. https://arxiv.org... on PithPith's reviewarXiv
- [2] Salaheddin Alzubi, N. Provenzano, Jaydon Bingham, Wei-Yuan Chen, and Tu Vu. 2026. https://arxiv.org/abs/2603.02766 EvoSkill: Automated Skill Discovery for Multi-Agent Systems . Preprint, arXiv:2603.02766 on PithPith's reviewarXiv
- [3] Minghao Chen, Yihang Li, Yan-Ting Yang, Shiyu Yu, Binbin Lin, and Xiaofei He. 2024. https://doi.org/10.52202/079017-0019 AutoManual: Constructing Instruction Manuals by LLM Agents via Interactive Environmental Learning . In Neural Information Processing Systems on Pithdoi:10.52202/079017-0019
- [4] Xin-Yun Chen, Maxwell Lin, Nathanael Sch \"a rli, and Denny Zhou. 2023. https://arxiv.org/abs/2304.05128 Teaching Large Language Models to Self-Debug . In International Conference on Learning Representations on PithPith's reviewarXiv
- [5] B. Efron. 1979. https://doi.org/10.1214/AOS/1176344552 Bootstrap Methods: Another Look at the Jackknife . Annals of Statistics, 7:1--26 on Pithdoi:10.1214/aos/1176344552
- [6] Yi-Chun Feng, Jia-Wei Wang, and Hao Sun. 2026. https://arxiv.org/abs/2609.29154 A Wrong Turn Does Not Ruin the Journey: Deviation-Guided Skill Self-Evolution for LLM Agents . Preprint, arXiv:2609.29154 on PithPith's reviewarXiv
- [7] Tingxu Han, Y. Zhang, Wei Song, Chun-Rong Fang, Zhen-Yu Chen, Youcheng Sun, and Li-Jie Hu. 2026. https://arxiv.org/abs/2603.15401 SWE-Skills-Bench: Do Agent Skills Actually Help in Real-World Software Engineering? Preprint, arXiv:2603.15401 on PithPith's reviewarXiv
- [8] James A. Jones and M. J. Harrold. 2005. https://doi.org/10.1145/1101908.1101949 Empirical evaluation of the tarantula automatic fault-localization technique . In International Conference on Automated Software Engineering on PithPith's reviewdoi:10.1145/1101908.1101949
- [9] Hao Kang and Ming Wen. 2026. https://arxiv.org/abs/2609.15396 SkillLift: Learning Dense Rubrics from Sparse Oracles for Efficient Skill Evolution . Preprint, arXiv:2609.15396 on PithPith's reviewarXiv
- [10] O. Khattab, Arnav Singhvi, Paridhi Maheshwari, Zhiyuan Zhang, Keshav Santhanam, Sri Vardhamanan, Saiful Haq, Ashutosh Sharma, Thomas T. Joshi, Hanna Moazam, Heather Miller, Matei Zaharia, and Christopher Potts. 2023. https://arxiv.org/abs/2310.03714 DSPy: Compiling Declarative... on PithPith's reviewarXiv
- [11] Ming-Hao Li. 2026. https://arxiv.org/abs/2609.30123 HEXIS: Compiling Agent Skills into Extended Finite State Machines . Preprint, arXiv:2609.30123 on PithPith's reviewarXiv
- [12] Xiangyi Li, Wen-Bo Chen, Yimin Liu, Sheng-Han Zheng, Xiaokun Chen, Yifeng He, Yubo Li, B. You, Haotian Shen, Jiankai Sun, Shuyi Wang, Qun-Hong Zeng, Di Wang, Xuandong Zhao, Xiaojun Wu, Yuan-Li Wang, Roey Ben Chaim, Zong-Lin Di, Yi Gao, and 23 others. 2026. https://arxiv.org/ab... on PithPith's reviewarXiv
- [13] Yuxuan Liu, Zhao-Chen Su, Lin Xie, Yuhao Zhang, Qing Zong, Jiahe Guo, Zhong-Wei Xie, Yiyan Ji, Yauwai Yim, Hong-Yu Luo, Xiyu Ren, Chenyu Ruan, Haoran Li, and Yang-Qiu Song. 2026. https://arxiv.org/abs/2606.01139 SkillRevise: Improving LLM-Authored Agent Skills via Trace-Condit... on PithPith's reviewarXiv
- [14] Aman Madaan, Niket Tandon, Prakhar Gupta, Skyler Hallinan, Luyu Gao, Sarah Wiegreffe, Uri Alon, Nouha Dziri, Shrimai Prabhumoye, Yiming Yang, Shashank Gupta, Bodhisattwa Prasad Majumder, Katherine Hermann, S. Welleck, A. Yazdanbakhsh, and Peter Clark. 2023. Self-Refine: Iterat... on Pith
- [15] Alistair Mavin, Philip Wilkinson, Adrian Harwood, and Mark Novak. 2009. https://doi.org/10.1109/RE.2009.9 Easy Approach to Requirements Syntax (EARS) . In IEEE International Requirements Engineering Conference on Pithdoi:10.1109/re.2009.9
- [16] Q. McNemar. 1947. https://doi.org/10.1007/BF02295996 Note on the sampling error of the difference between correlated proportions or percentages . Psychometrika, 12:153--157 on Pithdoi:10.1007/bf02295996
- [17] Yunjia Qi, Hao Peng, Xiaozhi Wang, Amy Xin, Youfeng Liu, Bin Xu, Lei Hou, and Juan-Zi Li. 2025. https://arxiv.org/abs/2505.16944 AGENTIF: Benchmarking Instruction Following of Large Language Models in Agentic Scenarios . Preprint, arXiv:2505.16944 on PithPith's reviewarXiv
- [18] Maksim Shaposhnikov, N. Fortuin, Simon Stipcich, Maria I. Gorinova, Amy Heineike, and Robert S. Willoughby. 2026. https://arxiv.org/abs/2606.17819 A Framework for Evaluating Agentic Skills at Scale . Preprint, arXiv:2606.17819 on PithPith's reviewarXiv
- [19] Noah Shinn, Federico Cassano, Edward Berman, A. Gopinath, Karthik Narasimhan, and Shunyu Yao. 2023. https://doi.org/10.52202/075280-0377 Reflexion: language agents with verbal reinforcement learning . In Neural Information Processing Systems on Pithdoi:10.52202/075280-0377
- [20] M. Suzgun, Mert Yuksekgonul, Federico Bianchi, Daniel Jurafsky, and James Zou. 2025. Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory . In Conference of the European Chapter of the Association for Computational Linguistics on Pith
show 14 more
- [21] Boyin Tan, Xiaowei Huang, and Youcheng Sun. 2026. https://doi.org/10.48550/arXiv.2606.20659 Skill Coverage : A test adequacy metric for agent skills . Preprint, arXiv:2606.20659 on PithPith's reviewarXivdoi:10.48550/arxiv.2606.20659
- [22] Guanzhi Wang, Yuqi Xie, Yunfan Jiang, A. Mandlekar, Chaowei Xiao, Yuke Zhu, Linxi (Jim) Fan, and Anima Anandkumar. 2023. https://arxiv.org/abs/2305.16291 Voyager: An Open-Ended Embodied Agent with Large Language Models . Transactions on Machine Learning Research on PithPith's reviewarXiv
- [23] Hao-Yu Wang, Christopher M. Poskitt, and Jun Sun. 2025. https://doi.org/10.1145/3744916.3764546 AgentSpec: Customizable Runtime Enforcement for Safe and Reliable LLM Agents . In Proceedings of the 2026 IEEE/ACM 48th International Conference on Software Engineering on Pithdoi:10.1145/3744916.3764546
- [24] Xingyao Wang, Boxuan Li, Yufan Song, Frank F. Xu, Xiangru Tang, Ming-Chen Zhuge, Jiayi Pan, Yue-Qi Song, Bo-Wen Li, Jaskirat Singh, Hoang H. Tran, Fuqiang Li, Ren Ma, Mingzhang Zheng, Bill Qian, Yanjun Shao, Niklas Muennighoff, Yi-Zhe Zhang, Binyuan Hui, and 5 others. 2024 a .... on Pith
- [25] Zora Zhiruo Wang, Jia-Yuan Mao, Daniel Fried, and Graham Neubig. 2024 b . Agent Workflow Memory . In International Conference on Machine Learning on Pith
- [26] W. E. Wong, Ruizhi Gao, Yi-hao Li, Rui Abreu, and Franz Wotawa. 2016. https://doi.org/10.1109/TSE.2016.2521368 A Survey on Software Fault Localization . IEEE Transactions on Software Engineering, 42:707--740 on Pithdoi:10.1109/tse.2016.2521368
- [27] Chengrun Yang, Xue-Zhi Wang, Yifeng Lu, Hanxiao Liu, Quoc V. Le, Denny Zhou, and Xin-Yun Chen. 2023. Large Language Models as Optimizers . In International Conference on Learning Representations on Pith
- [28] Mert Yuksekgonul, Federico Bianchi, Joseph Boen, Sheng Liu, Zhi Huang, Carlos Guestrin, and James Zou. 2024. https://arxiv.org/abs/2406.07496 TextGrad: Automatic "Differentiation" via Text . Preprint, arXiv:2406.07496 on PithPith's reviewarXiv
- [29] Han-Rong Zhang, Shi-Chen Fan, Henry Peng Zou, Yan-Kai Chen, Zhenting Wang, Jia-Yuan Zhou, Cheng-Ze Li, Wei-Chieh Huang, Yi-Fei Yao, Ke-Ning Zheng, Xue Liu, Xiao-Xiao Li, and Philip S. Yu. 2026. https://arxiv.org/abs/2604.01687 CoEvoSkills: Self-Evolving Agent Skills via Co-Evo... on PithPith's reviewarXiv
- [30] Qizheng Zhang, Chang-Ran Hu, Shubhangi Upasani, Bo-Yuan Ma, Feng-Lu Hong, Vamsidhar Kamanuru, Jay Rainton, Chen Wu, Meng-Meng Ji, Han-Chen Li, Urmish Thakker, James Zou, and K. Olukotun. 2025. https://arxiv.org/abs/2510.04618 Agentic Context Engineering: Evolving Contexts for ... on PithPith's reviewarXiv
- [31] Andrew Zhao, Daniel Huang, Quentin Xu, Matthieu Lin, Y. Liu, and Gao Huang. 2023. ExpeL: LLM Agents Are Experiential Learners . In AAAI Conference on Artificial Intelligence on Pith
- [32] Lian-Min Zheng, Wei-Lin Chiang, Ying Sheng, Si-Yuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuo-Han Li, Da-Cheng Li, E. Xing, Haotong Zhang, Joseph E. Gonzalez, and Ion Stoica. 2023. https://doi.org/10.52202/075280-2020 Judging LLM-as-a-judge with MT-Bench and Chatbot Ar... on Pithdoi:10.52202/075280-2020
- [33] Shan Zhong, Yi Lu, Jing-Jie Ning, Yi Wan, Li-Hang Feng, Y. Ao, Leonardo F. R. Ribeiro, Markus Dreyer, Sean Ammirati, and Chen-Yan Xiong. 2026. https://arxiv.org/abs/2604.20087 SkillLearnBench: Benchmarking Continual Learning Methods for Agent Skill Generation on Real-World Tas... on PithPith's reviewarXiv
- [34] Jeffrey Zhou, Tianjian Lu, Swaroop Mishra, Siddhartha Brahma, Sujoy Basu, Yi Luan, Denny Zhou, and Le Hou. 2023. https://arxiv.org/abs/2311.07911 Instruction-Following Evaluation for Large Language Models . Preprint, arXiv:2311.07911 on PithPith's reviewarXiv
Challenges and discussion
Disagree with the review, or know something it missed? Leave a comment, or challenge this review formally: a challenge becomes a permanent entry in this paper's record, and its decision is published here.
Come back by email
A short list of the week's newly reviewed papers. Nothing is sent until you confirm from your inbox, and every email has a link that stops it. No account.
No comments yet.
Sign in to comment