{"arxiv_id":"2606.15017","incoming":{"papers":1,"sentences":1,"groups":[],"unlabelled_sentences":1},"outgoing":{"papers":6,"sentences":10,"groups":[{"kind":"background","label":"Cites as background","papers":4,"sentences":4,"quotes":[{"arxiv_id":"2604.00073","title":"Terminal Agents Suffice for Enterprise Automation","url":"/paper/2604.00073","section":"Related Work","sentence":"A line of work explores structurally different paradigms: API-based agents substantially outperform browser agents on WebArena, with hybrid agents reaching even higher performance [ref] , and terminal-based agents operating through code and programmatic interfaces can match or outperform GUI-based agents on enterprise automation tasks [ref] .","intent_p":0.89,"contests_p":null},{"arxiv_id":"2604.13318","title":"WebXSkill: Skill Learning for Autonomous Web Agents","url":"/paper/2604.13318","section":"Related Work","sentence":"WALT [ref] learns stable tool abstractions reverse-engineered from website structure prior to evaluation, and WebXSkill [ref] mines parameterized action programs from synthetic agent trajectories.","intent_p":0.98,"contests_p":null},{"arxiv_id":"2504.07079","title":"SkillWeaver: Web Agents can Self-Improve by Discovering and Honing Skills","url":"/paper/2504.07079","section":"Related Work","sentence":"SkillWeaver [ref] discovers and synthesizes Python API skills in an offline phase before deployment.","intent_p":0.96,"contests_p":null},{"arxiv_id":"2510.04373","title":null,"url":"/paper/2510.04373","section":"Related Work","sentence":"JEF-Hinter [ref] distills offline trajectories into compact, context-aware hints, including from both successful and failed traces, and retrieves relevant hints for the agent's current state at inference time.","intent_p":0.8,"contests_p":null}]}],"unlabelled_sentences":6},"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/2606.15017"}}