# SkillGrad: Optimizing Agent Skills Like Gradient Descent

arXiv: https://arxiv.org/abs/2605.27760 · HTML: https://mlsanity.com/paper/2605.27760/

## Key facts

- Cited by 26 papers in the map (data through 14 Oct 2026).
- Placed in Agent skill learning and evolution, one of 16 directions in LLM agents and tool use.
- Builds on 16 of its 62 references that are in the map.
- First posted on arXiv 26 May 2026.

## Placed in

- LLM agents and tool use / [Agent skill learning and evolution](https://mlsanity.com/agents/agent-skill-learning-and-evolution/)

## Builds on (most cited first, 16 of 16)

- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Mind2Web: Towards a Generalist Agent for the Web](https://mlsanity.com/paper/2306.06070/) (2023)
- [Compositional Semantic Parsing on Semi-Structured Tables](https://mlsanity.com/paper/1508.00305/) (2015)
- [SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks](https://mlsanity.com/paper/2602.12670/) (2026)
- [Plan-and-Act: Improving Planning of Agents for Long-Horizon Tasks](https://mlsanity.com/paper/2503.09572/) (2025)
- [WebDancer: Towards Autonomous Information Seeking Agency](https://mlsanity.com/paper/2505.22648/) (2025)
- [Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills](https://mlsanity.com/paper/2603.25158/) (2026)
- [Assisting in Writing Wikipedia-like Articles From Scratch with Large Language Models](https://mlsanity.com/paper/2402.14207/) (2024)
- [CoEvoSkills: Self-Evolving Agent Skills via Co-Evolutionary Verification](https://mlsanity.com/paper/2604.01687/) (2026)
- [When MOE Meets LLMs: Parameter Efficient Fine-tuning for Multi-task Medical Applications](https://mlsanity.com/paper/2310.18339/) (2023)
- [SpreadsheetBench: Towards Challenging Real World Spreadsheet Manipulation](https://mlsanity.com/paper/2406.14991/) (2024)
- [SkillClaw: Let Skills Evolve Collectively with Agentic Evolver](https://mlsanity.com/paper/2604.08377/) (2026)
- [AgentGym-RL: Training LLM Agents for Long-Horizon Decision Making through Multi-Turn Reinforcement Learning](https://mlsanity.com/paper/2509.08755/) (2025)
- [MemCollab: Cross-Model Memory Collaboration via Contrastive Trajectory Distillation](https://mlsanity.com/paper/2603.23234/) (2026)
- [The Illusion of Reasoning: Exposing Evasive Data Contamination in LLMs via Zero-CoT Truncation](https://mlsanity.com/paper/2605.21856/) (2026)
- [Environment-in-the-Loop: Rethinking Code Migration with LLM-based Agents](https://mlsanity.com/paper/2602.09944/) (2026)

## Built on by (most cited first, 26 of 26)

- [WikiSkill: Compiling Agent Experience into Persistent Knowledge for Skill Evolution](https://mlsanity.com/paper/2608.27454/) (2026)
- [Skill Self-Play: Pushing the Frontier of LLM Capability with Co-Evolving Skills](https://mlsanity.com/paper/2607.22529/) (2026)
- [Dynamic Agent Skills: A Lifecycle Survey and Taxonomy of Evolving Skill Libraries](https://mlsanity.com/paper/2607.10113/) (2026)
- [SkillProx: Self-Evolving Agent Skills via Proximal Textual Gradient Descent](https://mlsanity.com/paper/2608.07449/) (2026)
- [SkillZip: Evaluation-Free Skill Compression for Self-Evolving Agents by Discovering Reusable Structure](https://mlsanity.com/paper/2608.11079/) (2026)
- [SKILL-KD: Contrastive Skill Distillation for LLM Agents](https://mlsanity.com/paper/2607.28048/) (2026)
- [Workflow-Localized Mechanism Learning: Attribution-Guided Repair and Knowledge Reuse for Structured Agent Skills](https://mlsanity.com/paper/2607.20999/) (2026)
- [Bayesian-Agent: Posterior-Guided Skill Evolution Across LLM Agent Harnesses](https://mlsanity.com/paper/2606.08348/) (2026)
- [Marginal Advantage Accumulation for Memory-Driven Agent Self-Evolution](https://mlsanity.com/paper/2606.20475/) (2026)
- [Rethinking Self-Evolution: A Constrained Exploration-Exploitation Process for Mitigating Skill Overfitting](https://mlsanity.com/paper/2607.26643/) (2026)
- [Do Personalized Skills Help Coding Agents? An Empirical Study of Developer Interaction Histories](https://mlsanity.com/paper/2608.10319/) (2026)
- [From Execution to Capability: Scientific Experience Consolidation via Procedural Knowledge Synthesis](https://mlsanity.com/paper/2607.24459/) (2026)
- [Skill Training with Corruption and Reconstruction Loop](https://mlsanity.com/paper/2607.27557/) (2026)
- [Branch2Skill: Efficient Skill Evolution Through Reasoning Trees](https://mlsanity.com/paper/2608.08677/) (2026)
- [MindMemOS: A Portable and Self-Evolving Memory Operating Layer for AI Agents](https://mlsanity.com/paper/2608.12428/) (2026)
- [CAPO: Constraint-Aware Prompt Optimization for LLM Agents](https://mlsanity.com/paper/2608.16068/) (2026)
- [Towards a Systems Foundation for Agentic Skills: Architecture, Lifecycle, and Security](https://mlsanity.com/paper/2608.29596/) (2026)
- [SkillZip Pro: Execution-Aware Dynamic Compression of Progressively Loaded Skills for Self-Evolving Agents](https://mlsanity.com/paper/2608.30785/) (2026)
- [SkillLift: Learning Dense Rubrics from Sparse Oracles for Efficient Skill Evolution](https://mlsanity.com/paper/2609.15396/) (2026)
- [SkillFocus: Evolving Agent Skills via Capability Decomposition](https://mlsanity.com/paper/2609.34397/) (2026)
- [Semantic Projection for Continual Self-Evolution of Language Agents](https://mlsanity.com/paper/2609.36626/) (2026)
- [Retrieval-Augmented Skill Optimization via Cross-Harness Adaptation](https://mlsanity.com/paper/2609.38024/) (2026)
- [Learning What to Investigate Next: Meta-Reasoning for Long-Horizon Research Agents](https://mlsanity.com/paper/2610.02525/) (2026)
- [EVISKILL: Grounding Skill Evolution in Replayable Evidence](https://mlsanity.com/paper/2610.05030/) (2026)
- [Trajectory-Guided Fault Localization for Agent Skill Evolution](https://mlsanity.com/paper/2610.11858/) (2026)
- [HarnessNet: Feedback-driven Hierarchical Joint Harness Evolution for Multi-agent Systems](https://mlsanity.com/paper/2610.15221/) (2026)

## Abstract (arXiv)

Agent skills provide a lightweight way to adapt LLM agents to specialized domains by storing reusable procedural knowledge in structured files. However, whether downloaded from third parties or self-generated, these skills are often unreliable, incomplete, or outdated. Existing skill-evolution methods often address these deficiencies through heuristic reflections without an explicit optimization formulation. In this paper, we propose SkillGrad, a gradient-descent-inspired framework for optimizing agent skills. SkillGrad treats the skill package as a structured parameter to optimize in a gradient descent fashion: task executions provide trajectory-level loss evidence, automatic diagnoses then provide text-based gradients that indicate the correction directions. To stabilize optimization across iterations, a momentum agent accumulates recurring diagnostic patterns into a persistent memory overlay. Finally, an LLM-based patcher executes the parameter update by applying layer-aware edits to the skill package. Evaluated on SpreadsheetBench Verified and WikiTableQuestions, SkillGrad consistently outperforms training-based skill evolution baselines across two backbone LLMs, improving over the strongest training-based baseline by 6.7 percentage points on average. Ablations further show that momentum and contrastive diagnosis both contribute to the final skill quality.