Harness evolution improves LLM agents by learning from execution trajectories, but existing experience- and skill-based methods are less effective on long-horizon tasks. As interactions grow, useful evidence can be buried by redundant or outdated context, making context management itself a key bottleneck. We introduce...
Wei-Yuan Li, Jing-Heng Xu, Ai-Li Chen et al.· 0 citations
Structured multi-agent workflows exchange intermediate messages whose content and form can reveal private state even when the final output is safe. We identify selection-channel leakage: after authorization fixes what may be released, a private-state-aware choice among semantically valid realizations creates an additio...
Jing-Heng Xu, Long-Ze Fan, Ze-Yuan Wang et al.· 0 citations
Open-ended reinforcement learning often relies on rubric-based rewards for tasks without directly verifiable answers. Yet the policy and reward system form a dynamic feedback loop: as the policy optimizes the current reward, an initially useful reward system may become unreliable due to reward hacking or reduced respon...
Weiyuan Li, Aili Chen, Xin-Tao Wang et al.· 0 citations
LLM agents increasingly rely on third-party skills, using natural-language descriptions for selection and instruction bodies for planning. This progressive-disclosure design exposes two sequential control points to untrusted publishers: a static skill may steer an otherwise correct task onto an unnecessarily costly tra...
Junliang Liu, Ruo-Yu Li, Wenxin Tang et al.· 0 citations
CapLease is introduced, an authorization-consumption layer that follows proposal- and authority-level defenses, binds an authenticated user confirmation to a canonical action, and enforces transactional Issue-Prepare-Commit transitions, which identifies durable authorization state, rather than token representation alon...
Jingheng Xu, Long-Ze Fan, Zeyuan Wang et al.· 2 citations
Minimum-Necessary Communication is introduced, a typed semantic-declassification protocol that selects a task-sufficient disclosure from an application-authored candidate family and binds it to explicit recipient, purpose, forwarding, lifetime, logging, and memory scopes.
Jingheng Xu, Longze Fan, Zeyuan Wang et al.· 0 citations
SWE-Bench ProMax is introduced, an expert-curated, multilingual code refactoring benchmark of 170 instances drawn from real commits across seven programming languages, which presents a meaningful and unsaturated challenge for current AI coding agents.
Yuling Shi, Jingheng Xu, Kelin Fu et al.· 5 citations
HarnessLens is introduced, a budget-aware framework for automated harness evolution that jointly explores the task space and user-configurable components, derives candidate modifications from execution trajectories, and selectively verifies each candidate on behavior-relevant tasks using an attributable-evidence gate.
Jingheng Xu, Yi-Kai Zhang, Aiden Chen et al.· 2 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
