AFTER benchmark shows single refinement improves LLM agent performance by 3.7-6.7 points and multi-model procedural skills reach 73.1% cross-model accuracy on 382 tasks.
Skillreducer: Optimizing llm agent skills for token efficiency.arXiv preprint arXiv:2603.29919
12 Pith papers cite this work. Polarity classification is still indexing.
12
Pith papers citing it
abstract
LLM-based coding agents rely on \emph{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 \textsc{SkillReducer}, 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, \textsc{SkillReducer} achieves 48\% description compression and 39\% body compression while improving functional quality by 2.8\%, revealing a \emph{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.
years
2026 12verdicts
UNVERDICTED 12representative citing papers
Attacks can corrupt the latent future trajectory imagined by world-action models in VLA policies, causing failures in oracles like MPC while the reactive policy stays intact.
SkillAxe is an unsupervised framework that decomposes LLM skill quality into four dimensions to generate improvement briefs, raising pass rates 28% relative on SkillsBench and from 16% to 52% on SpreadsheetBench.
SkillPager retrieves typed semantic nodes from skill documents via MMR to reach 78.89% LLM-judged sufficiency with 47% fewer tokens than full documents on a 395-skill benchmark.
Skill shadowing, not context overhead, is the dominant cause of performance degradation when LLM agent skill libraries expand.
SkillMaster enables LLM agents to autonomously develop skills via trajectory review, counterfactual evaluation, and DualAdv-GRPO training, boosting success rates by 8.8% on ALFWorld and 9.3% on WebShop.
ObjectGraph is a Markdown superset file format that represents documents as traversable knowledge graphs, achieving up to 95.3% token reduction for agents with no significant accuracy loss.
OpenSkill bootstraps LLM agent self-evolution by pulling grounded knowledge and anchors from open-world sources, synthesizing transferable skills, and refining them on self-generated virtual tasks, achieving top benchmark pass rates without supervision.
ANX introduces a protocol-first design with 3EX architecture that cuts token consumption by 47-66% and execution time by 58% versus prior methods in form-filling tests.
SkillsInjector uses adaptive skill selection and set-aware rendering to improve LLM agent performance by 3.9-7.3 percentage points over baselines on three benchmarks.
Single-project case study identifies Index Sickness from complex symbolic LLM management and reports that Baseline-Log Physical Separation reduced instructions by 75% with no recurrence observed.
Data-centric optimization of skills for agents on a branching lakehouse improves accuracy by 31.9% on 25 tasks via state-verification evaluation.
citing papers explorer
-
Managing Procedural Memory in LLM Agents: Control, Adaptation, and Evaluation
AFTER benchmark shows single refinement improves LLM agent performance by 3.7-6.7 points and multi-model procedural skills reach 73.1% cross-model accuracy on 382 tasks.
-
Attacking the Trusted Imagination: Oracle-Level Integrity Attacks on Imagine-then-Act World Models
Attacks can corrupt the latent future trajectory imagined by world-action models in VLA policies, causing failures in oracles like MPC while the reactive policy stays intact.
-
SkillAxe: Sharpening LLM-Authored Agent Skills Through Evaluation-Guided Self-Refinement
SkillAxe is an unsupervised framework that decomposes LLM skill quality into four dimensions to generate improvement briefs, raising pass rates 28% relative on SkillsBench and from 16% to 52% on SpreadsheetBench.
-
SkillPager: Query-Adaptive Intra-Skill Navigation via Semantic Node Retrieval
SkillPager retrieves typed semantic nodes from skill documents via MMR to reach 78.89% LLM-judged sufficiency with 47% fewer tokens than full documents on a 395-skill benchmark.
-
More Skills, Worse Agents? Skill Shadowing Degrades Performance When Expanding Skill Libraries
Skill shadowing, not context overhead, is the dominant cause of performance degradation when LLM agent skill libraries expand.
-
SkillMaster: Toward Autonomous Skill Mastery in LLM Agents
SkillMaster enables LLM agents to autonomously develop skills via trajectory review, counterfactual evaluation, and DualAdv-GRPO training, boosting success rates by 8.8% on ALFWorld and 9.3% on WebShop.
-
ObjectGraph: From Document Injection to Knowledge Traversal -- A Native File Format for the Agentic Era
ObjectGraph is a Markdown superset file format that represents documents as traversable knowledge graphs, achieving up to 95.3% token reduction for agents with no significant accuracy loss.
-
OpenSkill: Open-World Self-Evolution for LLM Agents
OpenSkill bootstraps LLM agent self-evolution by pulling grounded knowledge and anchors from open-world sources, synthesizing transferable skills, and refining them on self-generated virtual tasks, achieving top benchmark pass rates without supervision.
-
ANX: Protocol-First Design for AI Agent Interaction with a Supporting 3EX Decoupled Architecture
ANX introduces a protocol-first design with 3EX architecture that cuts token consumption by 47-66% and execution time by 58% versus prior methods in form-filling tests.
-
SkillsInjector: Dynamic Skill Context Construction for LLM Agents
SkillsInjector uses adaptive skill selection and set-aware rendering to improve LLM agent performance by 3.9-7.3 percentage points over baselines on three benchmarks.
-
Written by AI, Managed by AI: Semantic Space Control and Index Sickness Elimination Across 391 Consecutive Sessions
Single-project case study identifies Index Sickness from complex symbolic LLM management and reports that Baseline-Log Physical Separation reduced instructions by 75% with no recurrence observed.
-
"Skill issues'': data-centric optimization of lakehouse agents
Data-centric optimization of skills for agents on a branching lakehouse improves accuracy by 31.9% on 25 tasks via state-verification evaluation.