{"arxiv_id":"2603.15401","incoming":{"papers":48,"sentences":81,"groups":[{"kind":"method_or_tool","label":"Uses its method or tool","papers":1,"sentences":1,"quotes":[{"arxiv_id":"2605.11665","title":"Nautilus: From One Prompt to Plug-and-Play Robot Learning","url":"/paper/2605.11665","section":"Ablation: from pairwise wiring to reusable extending","sentence":"Following [ref] on evaluating coding-agent performance, we measure and report: Wrap success ($↑$) measures whether the agent's generated integration loads the released checkpoint and completes the benchmark evaluation protocol end-to-end without manual intervention.","intent_p":0.88,"contests_p":null}]},{"kind":"data_used","label":"Uses its data or values","papers":2,"sentences":2,"quotes":[{"arxiv_id":"2605.10990","title":"Skill Drift Is Contract Violation: Proactive Maintenance for LLM Agent Skill Libraries","url":"/paper/2605.10990","section":": A Benchmark for Skill Degradation","sentence":"Controlled drift construction and validation. starts from 49 agent skills in SWE-Skills-Bench [ref] across seven programming domains.","intent_p":0.98,"contests_p":null},{"arxiv_id":"2608.11888","title":"Agent Skills Can Be Harmful: An Empirical Study of Skill-Induced Failures in LLM Agents","url":"/paper/2608.11888","section":"Study Subjects","sentence":"SWE-Skills-Bench provides 490 repository-based software-engineering task instances with no-skill and curated-skill settings [ref] .","intent_p":0.86,"contests_p":null}]},{"kind":"supports_this_paper","label":"Cites it as support","papers":5,"sentences":6,"quotes":[{"arxiv_id":"2607.10113","title":"Dynamic Agent Skills: A Lifecycle Survey and Taxonomy of Evolving Skill Libraries","url":"/paper/2607.10113","section":"Focused libraries often beat comprehensive ones","sentence":"SkillMOO [ref] optimizes pass rate and cost jointly, SkillLearnBench [ref] shows that plausible skill text does not imply competence, SWE-Skills-Bench [ref] reports only $+1.2%$ average gain across public SWE skills with many zero-utility or negative cases, and SkillFlow-Bench [ref] finds that stronger settings maintain compact revised skills while weaker settings often fragment into many low-utility files.","intent_p":0.97,"contests_p":null},{"arxiv_id":"2608.25399","title":"Can your AI agent be cheaper? Investigating the effects of task specifications on token spend in agentic coding tasks","url":"/paper/2608.25399","section":"Related work","sentence":"The distinction matters because task-relevant information is not monotonically beneficial: removing detail can improve correctness by disrupting misleading lexical cues [ref] , and packaged skill documents may raise token spend without improving performance [ref] .","intent_p":0.99,"contests_p":null},{"arxiv_id":"2608.20485","title":"Terminal Agents: A Survey of AI Agents in Command-Line Environments","url":"/paper/2608.20485","section":"Synthesis: Architectural Trade-offs and Attribution Consequences","sentence":"Controlled skill injection produces task-dependent gains and can add substantial token overhead without improving pass rate [ref] .","intent_p":0.98,"contests_p":null},{"arxiv_id":"2607.06283","title":"Task Decomposition-Guided Reranking for Adaptive Agent Skill Retrieval","url":"/paper/2607.06283","section":"Introduction","sentence":"Selecting too few skills may be insufficient to support the completion of complex tasks, while selecting too many may introduce redundant information, increase the agent's decision-making burden, and even interfere with subsequent execution [ref] .","intent_p":0.96,"contests_p":null},{"arxiv_id":"2607.01248","title":"A Practice Auditing Framework for Large Language Model Use: Collective Empiricism, Pseudo-Rational Cognition, and Governance of AI-Generated Content","url":"/paper/2607.01248","section":"AI Skill Summarization and Skill Debt","sentence":"The effectiveness of a skill depends on usage frequency, scenario relevance, task association, failure records, invocation boundaries, and human feedback [ref] .","intent_p":0.97,"contests_p":null}]},{"kind":"background","label":"Cites it as background","papers":18,"sentences":22,"quotes":[{"arxiv_id":"2604.09297","title":"SkillMOO: Multi-Objective Optimization of Agent Skills for Software Engineering","url":"/paper/2604.09297","section":"Introduction","sentence":"Related Work SkillsBench [ref] introduced the first systematic paired evaluation of skills across diverse tasks, and SWE-Skills-Bench [ref] found that many public SWE skills provide little benefit in practical SE tasks.","intent_p":0.89,"contests_p":null},{"arxiv_id":"2608.14036","title":"Demystifying Agent Skills: Why They Work-Until They Don't","url":"/paper/2608.14036","section":"Benchmarks for Evaluating Agent Behavior","sentence":"SkillsBench [ref] measures structured skills across diverse tasks, while SWE-Skills-Bench [ref] studies the marginal utility of skill documents in real-world software-engineering settings.","intent_p":0.83,"contests_p":null},{"arxiv_id":"2608.10319","title":"Do Personalized Skills Help Coding Agents? An Empirical Study of Developer Interaction Traces","url":"/paper/2608.10319","section":"Introduction","sentence":"Agent skills have recently emerged as a promising mechanism for retaining and transferring experience, by packaging reusable procedural knowledge as structured instructions for LLM-based agents [ref] .","intent_p":0.93,"contests_p":null},{"arxiv_id":"2605.10990","title":"Skill Drift Is Contract Violation: Proactive Maintenance for LLM Agent Skill Libraries","url":"/paper/2605.10990","section":"Related Work","sentence":"Agent benchmarks such as AgentBench [ref] , SWE-bench [ref] , and SWE-Skills-Bench [ref] evaluate task-solving, software engineering, and skill quality.","intent_p":0.98,"contests_p":null},{"arxiv_id":"2605.23899","title":"From Raw Experience to Skill Consumption: A Systematic Study of Model-Generated Agent Skills","url":"/paper/2605.23899","section":"Related Work","sentence":"One group focuses on whether skills help at all: SkillsBench [ref] , SWE-Skills-Bench [ref] , and [ref] primarily test whether curated or discovered skills improve downstream performance over a no-skill baseline.","intent_p":0.82,"contests_p":null}]}],"unlabelled_sentences":50},"outgoing":{"papers":0,"sentences":0,"groups":[],"unlabelled_sentences":0},"claims":[],"thresholds":{"intent_p":0.8,"supports_p":0.95,"contests_p":0.7},"method":"Citing sentences come from each paper's LaTeX source; every \\cite marks the sentence and the works it cites. Pith's classifier typed each sentence. A label is shown at intent probability 0.8 or more (0.95 for support), a contest at 0.7 or more. Measured accuracy against two blind model families is at https://pith.science/accuracy. The quoted sentences are the citing authors' own.","held_up":null,"_links":{"html":"https://pith.science/paper/2603.15401"}}