{"arxiv_id":"2608.09253","incoming":{"papers":0,"sentences":0,"groups":[],"unlabelled_sentences":0},"outgoing":{"papers":10,"sentences":18,"groups":[{"kind":"method_or_tool","label":"Uses the method or tool of","papers":1,"sentences":1,"quotes":[{"arxiv_id":"2605.13044","title":"No Attack Required: Semantic Fuzzing for Specification Violations in Agent Skills","url":"/paper/2605.13044","section":"Evaluation Setup","sentence":"Since each retained SkillsBench skill originally provides only one task query, we expand the query set for each selected skill through a two-stage augmentation process inspired by [ref] .","intent_p":0.97,"contests_p":null}]},{"kind":"data_used","label":"Uses data or values from","papers":1,"sentences":1,"quotes":[{"arxiv_id":"2602.12670","title":"SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks","url":"/paper/2602.12670","section":"Evaluation Setup","sentence":"We build our evaluation dataset based on SkillsBench [ref] , which contains 87 tasks associated with one user query, a skill document, a task execution environment and an outcome-based deterministic verifier.","intent_p":0.96,"contests_p":null}]},{"kind":"supports_this_paper","label":"Cites as support","papers":1,"sentences":1,"quotes":[{"arxiv_id":"2601.06112","title":null,"url":"/paper/2601.06112","section":"Introduction","sentence":"Even when the LLM agent has demonstrated the capability to complete a task under the guidance of a skill before, it may fail to reproduce the correct execution consistently in later runs [ref] .","intent_p":0.98,"contests_p":null}]},{"kind":"background","label":"Cites as background","papers":6,"sentences":13,"quotes":[{"arxiv_id":"2603.25158","title":"Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills","url":"/paper/2603.25158","section":"Introduction","sentence":"Some automatic prompt optimization and reflection-based approaches refine prompts, instructions, in-context memories, or skills based on execution feedback [ref] .","intent_p":0.89,"contests_p":null},{"arxiv_id":"2604.20087","title":"SkillLearnBench: Benchmarking Continual Learning Methods for Agent Skill Generation on Real-World Tasks","url":"/paper/2604.20087","section":"Background","sentence":"To guide multi-step planning and execution, skills have been introduced as reusable procedural knowledge for completing a class of tasks [ref] .","intent_p":0.99,"contests_p":null},{"arxiv_id":"2602.12670","title":"SkillsBench: Benchmarking How Well Agent Skills Work Across Diverse Tasks","url":"/paper/2602.12670","section":"Background","sentence":"To guide multi-step planning and execution, skills have been introduced as reusable procedural knowledge for completing a class of tasks [ref] .","intent_p":0.99,"contests_p":null},{"arxiv_id":"2605.18693","title":"SkillGenBench: Benchmarking Skill Generation Pipelines for LLM Agents","url":"/paper/2605.18693","section":"Background","sentence":"To guide multi-step planning and execution, skills have been introduced as reusable procedural knowledge for completing a class of tasks [ref] .","intent_p":0.99,"contests_p":null},{"arxiv_id":"2210.03629","title":"ReAct: Synergizing Reasoning and Acting in Language Models","url":"/paper/2210.03629","section":"Background","sentence":"An LLM agent completes tasks through iterative interactions with external environments [ref] .","intent_p":1.0,"contests_p":null}]}],"unlabelled_sentences":2},"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/2608.09253"}}