🤖 AI Agent 研究Research
GUI代理在动态图形用户界面上执行长时间任务，其中弹出窗口、延迟加载和重新定位的小部件通常会使执行前修复的计划无效。我们提出了EvoSkill-GUI ，这是一个无需培训的框架，其中每个技能都是一个结构化的多文件包，包含检索元数据、可执行计划、备份本地化、故障恢复规则、可访问性实用程序和故障情况。
GUI agents execute long-horizon tasks on dynamic graphical user interfaces, where pop-ups, delayed loads, and relocated widgets routinely invalidate plans fixed before execution. We propose EvoSkill-GUI, a training-free framework in which each skill is a structured multi-file package containing retrieval metadata, executable plans, backup localization, failure-recovery rules, accessibility utilities, and failure cases.
科学代理人通过综合证据、评估建议和开发新的解释来为假设的发现做出贡献。在此基础上，我们引入了HypoEvolve ，它通过对假设群体的连续更新来明确协作。
Scientific agents contribute to hypothesis discovery by synthesizing evidence, assessing proposals, and developing new explanations. Building on this view, we introduce HypoEvolve, which makes collaboration explicit through successive updates to a hypothesis population.
大型语言模型（ LLM ）交易代理可以结合市场数据、新闻和可执行文件分析，但其行为通常由部署前固定的静态手写工具使用策略控制。我们引入了EvolveTrade ，这是一个自我发展的框架，将使用工具的交易代理的系统提示视为文本参数化策略。
Large language model (LLM) trading agents can combine market data, news, and executable analysis, but their behavior is often controlled by static hand-written tool-use policies that are fixed before deployment. We introduce EvolveTrade, a self-evolving framework that treats the system prompt of a tool-using trading agent as a text-parameterized policy.
科学代码库在可执行模型、方法和工具中编码了数十年的人类知识。我们引入了ScienceIDE ，这是将世界科学代码转化为科学代理可编程环境的基础设施。
Scientific code repositories encode decades of human knowledge in executable models, methods, and tools. We introduce ScienceIDE, infrastructure for turning the world’s scientific code into programmable environments for scientific agents.
⭐ GitHub 热门项目GitHub Trending
【GitHub】让任何AI代理通过Apple快捷方式读取您的Mac —日历、提醒、备注、联系人、剪贴板、位置、天气。只读，本地，无账号。（⭐ 8 ）
【GitHub】Let any AI agent read your Mac — Calendar, Reminders, Notes, Contacts, Clipboard, Location, Weather — through Apple Shortcuts. Read-only, local, no account. （⭐ 8）
【GitHub】AI智能体的自适应推理调度—每轮模型、工作量和编码线束的提供商路由：命令代码mod或本地OpenAI兼容的代理。（⭐ 9 ）
【GitHub】Adaptive inference scheduling for AI agents — per-round model, effort and provider routing for coding harnesses: a Command Code mod or a local OpenAI-compatible proxy. （⭐ 9）
【GitHub】受Claude Code启发的⚡免费开源终端编码助手，由Google Gemini提供支持。具备俄罗斯智能代理、实时配额HUD、自主子代理功能。（⭐ 0 ）
【GitHub】⚡ Free, open-source terminal coding assistant inspired by Claude Code, powered by Google Gemini. Featuring smart proxying for Russia, real-time quota HUD, and autonomous subagents. （⭐ 0）
【GitHub】使用Voidtools Everything为AI代理（ Cursor、Codex、PI-Desktop ）提供⚡超快速、索引驱动的本地文件搜索技能（⭐ 43 ）
【GitHub】⚡ Ultra-fast, index-powered local file search skill for AI agents (Cursor, Codex, PI-Desktop) using Voidtools Everything （⭐ 43）
【GitHub】Sno Station —您的Claude Code和Codex在您自己的机器上作为一个小队工作。共享加密内存、座席到座席消息传递（ Reach ）、交接和跨供应商审核的小队技能，以及ni （⭐ 0 ）
【GitHub】Sno Station — your Claude Code and Codex working as one squad on your own machine. Shared encrypted memory, agent-to-agent messaging (Reach), squad skills for handoff and cross-vendor review, and a ni （⭐ 0）
🚀 模型与行业动态Models & Industry
特朗普在没有证据的情况下声称，人工智能的反弹是民主党的骗局。特朗普表示，现在是时候用一个新名字重塑人工智能了，他还在创建一支人工智能部队。
Trump claimed, without evidence, that the AI backlash is a Democratic hoax.. Trump says it’s time to rebrand AI with a new name — and he’s also creating an AI Force.
谷歌表示， Gemini通过立即终止每次黑客攻击“采取了适当的行动”。谷歌的Gemini是攻击其他公司的最新人工智能模型。
Google said Gemini had "acted appropriately" by ending each hack immediately.. Google’s Gemini is the latest AI model to hack other companies.
Petlibro的新型Granary 2智能喂食器使用内置比例和（在价格较高的型号上）人工智能摄像头来准确跟踪您的猫吃多少以及何时吃—尽管最昂贵的健康监测功能将花费您额外的订阅费用。
Petlibro's new Granary 2 smart feeders use a built-in scale and (on pricier models) an AI camera to track exactly how much your cat is eating and when — though the fanciest health-monitoring features will cost you an extra subscription.
本周，两场关于人工智能安全的对话如火如荼，表明从虚构中辨别人工智能事实是多么困难。
This week two conversations about AI safety went viral that demonstrate just how hard it is to discern AI fact from fiction.
🔥 社区热议Community
【HN】热度: 1094 分 | 268 评论
【HN】热度: 1094 分 | 268 评论
【HN】热度: 129 分 | 36 评论
【HN】热度: 129 分 | 36 评论