# SkillReducer: Optimizing LLM Agent Skills for Token Efficiency

arXiv: https://arxiv.org/abs/2603.29919 · HTML: https://mlsanity.com/paper/2603.29919/

## Key facts

- Cited by 33 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 24 of its 33 references that are in the map.
- First posted on arXiv 31 Mar 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, 24 of 24)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [ToolLLM: Facilitating Large Language Models to Master 16000+ Real-world APIs](https://mlsanity.com/paper/2307.16789/) (2023)
- [AgentBench: Evaluating LLMs as Agents](https://mlsanity.com/paper/2308.03688/) (2023)
- [ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools](https://mlsanity.com/paper/2406.12793/) (2024)
- [Large Language Models Are Human-Level Prompt Engineers](https://mlsanity.com/paper/2211.01910/) (2022)
- [Large Language Models Can Be Easily Distracted by Irrelevant Context](https://mlsanity.com/paper/2302.00093/) (2023)
- [API-Bank: A Comprehensive Benchmark for Tool-Augmented LLMs](https://mlsanity.com/paper/2304.08244/) (2023)
- [LongLLMLingua: Accelerating and Enhancing LLMs in Long Context Scenarios via Prompt Compression](https://mlsanity.com/paper/2310.06839/) (2023)
- [Understanding the planning of LLM agents: A survey](https://mlsanity.com/paper/2402.02716/) (2024)
- [Learning to Compress Prompts with Gist Tokens](https://mlsanity.com/paper/2304.08467/) (2023)
- [SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks](https://mlsanity.com/paper/2602.12670/) (2026)
- [RECOMP: Improving Retrieval-Augmented LMs with Compression and Selective Augmentation](https://mlsanity.com/paper/2310.04408/) (2023)
- [Compressing Context to Enhance Inference Efficiency of Large Language Models](https://mlsanity.com/paper/2310.06201/) (2023)
- [Context Length Alone Hurts LLM Performance Despite Perfect Retrieval](https://mlsanity.com/paper/2510.05381/) (2025)
- [Agent Skills in the Wild: An Empirical Study of Security Vulnerabilities at Scale](https://mlsanity.com/paper/2601.10338/) (2026)
- [Prompt Compression for Large Language Models: A Survey](https://mlsanity.com/paper/2410.12388/) (2024)
- [Speed at the Cost of Quality: How Cursor AI Increases Short-Term Velocity and Long-Term Complexity in Open-Source Projects](https://mlsanity.com/paper/2511.04427/) (2025)
- [Concise and Precise Context Compression for Tool-Using Language Models](https://mlsanity.com/paper/2407.02043/) (2024)
- [ChatDBG: Augmenting Debugging with Large Language Models](https://mlsanity.com/paper/2403.16354/) (2024)
- [Debloating Software through Piece-Wise Compilation and Loading](https://mlsanity.com/paper/1802.00759/) (2018)
- [AgentCompress: Task-Aware Compression for Affordable Large Language Model Agents](https://mlsanity.com/paper/2601.05191/) (2026)

## Built on by (most cited first, 30 of 33)

- [SkillRevise: Improving LLM-Authored Agent Skills via Trace-Conditioned Skill Revision](https://mlsanity.com/paper/2606.01139/) (2026)
- [More Skills, Worse Agents? Skill Shadowing Degrades Performance When Expanding Skill Libraries](https://mlsanity.com/paper/2605.24050/) (2026)
- [OpenSkill: Open-World Self-Evolution for LLM Agents](https://mlsanity.com/paper/2606.06741/) (2026)
- [SkillAxe: Sharpening LLM-Authored Agent Skills Through Evaluation-Guided Self-Refinement](https://mlsanity.com/paper/2606.10546/) (2026)
- [Skill Self-Play: Pushing the Frontier of LLM Capability with Co-Evolving Skills](https://mlsanity.com/paper/2607.22529/) (2026)
- [SkillMaster: Toward Autonomous Skill Mastery in LLM Agents](https://mlsanity.com/paper/2605.08693/) (2026)
- [Agent Skill Evaluation and Evolution: Frameworks and Benchmarks](https://mlsanity.com/paper/2606.11435/) (2026)
- [Managing Procedural Memory in LLM Agents: Control, Adaptation, and Evaluation](https://mlsanity.com/paper/2606.23127/) (2026)
- [From Anatomy to Smells: An Empirical Study of SKILL.md in Agent Skills](https://mlsanity.com/paper/2607.01456/) (2026)
- [SkillZip: Evaluation-Free Skill Compression for Self-Evolving Agents by Discovering Reusable Structure](https://mlsanity.com/paper/2608.11079/) (2026)
- [Attacking the Trusted Imagination: Oracle-Level Integrity Attacks on Imagine-then-Act World Models](https://mlsanity.com/paper/2606.22966/) (2026)
- [SKILL-KD: Contrastive Skill Distillation for LLM Agents](https://mlsanity.com/paper/2607.28048/) (2026)
- [Ace-Skill: Bootstrapping Multimodal Agents with Prioritized and Clustered Evolution](https://mlsanity.com/paper/2605.08887/) (2026)
- [From Relevance to Execution Utility: Reward-Aware Dynamic Execution Gating for Skill-Based LLM Agents](https://mlsanity.com/paper/2608.09168/) (2026)
- [Optimal Skill Selection for LLM Agents with Provable Bicriteria Guarantees](https://mlsanity.com/paper/2608.19993/) (2026)
- [ObjectGraph: From Document Injection to Knowledge Traversal -- A Native File Format for the Agentic Era](https://mlsanity.com/paper/2604.27820/) (2026)
- ["Skill Issues'': Data-Centric Optimization of Lakehouse Agents](https://mlsanity.com/paper/2606.01185/) (2026)
- [SkillZip: Contract-Preserving Graph Compression for Scalable Agent Skill Libraries](https://mlsanity.com/paper/2608.05604/) (2026)
- [What Keeps Agent Skills from Being Reusable? Evidence from 138K SKILL.md Files](https://mlsanity.com/paper/2608.08453/) (2026)
- [SkillGate: Training In-Policy Skill Selection in Long-Horizon Agents](https://mlsanity.com/paper/2608.18852/) (2026)
- [The Router Within: Eliciting Native Skill Routing from a Frozen LLM](https://mlsanity.com/paper/2609.15982/) (2026)
- [Safe Skill Retirement for Physical Agents](https://mlsanity.com/paper/2609.29543/) (2026)
- [SkillSeek: Revisiting Agent Skill Retrieval at Marketplace Scale](https://mlsanity.com/paper/2609.38822/) (2026)
- [SkillPager: Query-Adaptive Intra-Skill Navigation via Semantic Node Retrieval](https://mlsanity.com/paper/2606.00822/) (2026)
- [Compression, structure, and executor capability: a controlled real-cost decomposition of language-model agent skill optimisation](https://mlsanity.com/paper/2607.03048/) (2026)
- [VCE-Skill: Enhancing Skill Self-Evolution with Version-Change Experience](https://mlsanity.com/paper/2608.16544/) (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)
- [Demystifying Agent Skills for Smart Contract Auditing: Design, Effectiveness, Behavioral Impact](https://mlsanity.com/paper/2609.29454/) (2026)
- [EVISKILL: Grounding Skill Evolution in Replayable Evidence](https://mlsanity.com/paper/2610.05030/) (2026)

## Abstract (arXiv)

LLM-based coding agents rely on skills, pre-packaged instruction sets that extend agent capabilities, yet every token of skill content injected into the context window incurs both monetary cost and attention dilution. To understand the severity of this problem, we conduct a large-scale empirical study of 55,315 publicly available skills and find systemic inefficiencies: 26.4% lack routing descriptions entirely, over 60% of body content is non-actionable, and reference files can inject tens of thousands of tokens per invocation. Motivated by these findings, we present textscSkillReducer, a two-stage optimization framework. Stage~1 optimizes the routing layer by compressing verbose descriptions and generating missing ones via adversarial delta debugging. Stage~2 restructures skill bodies through taxonomy-driven classification and progressive disclosure, separating actionable core rules from supplementary content loaded on demand, validated by faithfulness checks and a self-correcting feedback loop. Evaluated on 600 skills and the SkillsBench benchmark, textscSkillReducer achieves 48% description compression and 39% body compression while improving functional quality by 2.8%, revealing a less-is-more effect where removing non-essential content reduces distraction in the context window. These benefits transfer across five models from four families with a mean retention of 0.965, and generalize to an independent agent framework.