智能体生态观察
智能体生态动态速递
追踪 AI 智能体（Agent）生态的模型发布、框架开源、行业落地与安全治理动态。
Building a Streaming Robotics Learning Pipeline Using NVIDIA Cosmos3-DROID
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
Can AI Agents Make Open-Ended Scientific Discovery? Evidence from Station
Recent AI systems have made rapid progress in scientific discovery when given well-defined metrics, but whether they can…
Attacca: Goal-Directed Control under State Continuity for Long-Horizon Embodied Agents
A central capability of embodied agents is to accomplish complex objectives through sequences of interdependent tasks. Y…
Rethinking Cross-Tokenizer On-Policy Distillation: From Alignment Coverage to Supervision Reliability
On-Policy Distillation (OPD) trains a student on its own generations using teacher feedback. With different tokenizers, …
Reflection AI Introduces Beam: A 501B Open-Weight MoE Model With 23B Active Parameters for Coding and Agentic Workloads
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World Model
Web agents complete user requests by reading and acting on pages that third parties write, so an instruction planted on …
A GPU-Parallel Framework for Heterogeneous Multi-Task Reinforcement Learning
GPU-parallel simulation provides abundant robot interaction, but existing benchmarks rarely combine this scale with hete…
NeMo-DCR: Bit-Exact Delta-Compressed Refit for Scalable Agentic RL at Trillion-Parameter Scale
Agentic reinforcement learning (RL) disaggregates training from rollout, so each policy update must reach the rollout cl…
DAEDALUS: Bootstrapping Agent Memory from Self-Generated Tasks
LLM agents often lack the operational knowledge to act reliably in new environments, as they must discover specific tool…
Meet Together Link: A Free CLI That Runs Open Models Like Kimi K3 and GLM 5.3 Inside Claude Code, Codex, and OpenCode
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
Towards In-Parameter Memory Augmentation for Large Language Models
Recently Large Language Models (LLMs) and LLM-based agents increasingly need to incorporate knowledge acquired after pre…
From Evidence to Action: How Tool-Using Agents Fail
Tool-using agents make consequential changes to external state, yet correct outcomes do not guarantee that their actions…
Stateless Language Agents: Scaling Long-Horizon Automated Research
Automated research systems increasingly run LLM agents over long horizons, but more inference does not by itself produce…
Reka Releases Rho-1: A 19B Omni-Reasoning Model That Understands, Generates Video and Outputs Robot Actions in One
!(https://www.gstatic.com/images/branding/googleg/1x/googleg_standard_color_128dp.png)Add as a preferredsource on Google…
Coding Agent 是怎么工作的：规划、执行与验证的循环
Coding Agent 的核心是一个 ReAct 式循环：规划、执行、观察、验证、修正。本文拆解循环的工程实现——上下文管理、验证闭环、权限分级与三种典型失败模式，并说明 harness 与模型同样重要。
Kubernetes 官方沙箱项目 agent-sandbox 全解：CRD 模型、预热池、休眠机制与生产接入实施
K8s 官方 SIG Apps 沙箱编排项目的完整拆解与落地指南：Sandbox/Template/WarmPool/Claim 四件套的 CRD 模型与认领数据流、休眠与到期回收的生命周期设计、默认拒绝的托管网络策略与 Sandbox Router 原理，附预热池实测性能账本（突发 300 claims/s @ p90≤200ms）、调优旋钮与从安装、模板化、SDK 接入到生产检查清单的全流程实施路径。
Anthropic 将 Claude Cowork 任务迁移至云端：Pro/Max 用户 10 月 6 日起生效
10 月 6 日起，Pro 和 Max 计划的 Cowork 新任务改为在 Anthropic 云端运行，本地执行选项被移除，已启动的本地任务不受影响。
从补全到结对编程：AI 编程工具的三次跃迁
AI 编程工具经历了光标补全、对话式多文件编辑、自主 Coding Agent 三次跃迁。本文梳理每次跃迁的形态变化与背后「上下文工程」「工具调用」两个技术变量，并讨论工作流的改变与必须由人守住的边界。
给 Agent 设计工具的七条原则
Agent 表现的上限常常不在模型而在工具。本文总结七条工具设计原则——少而精、参数扁平化、描述写给模型看、返回值可行动、副作用显式化、幂等与超时、正交覆盖——多数配好坏对比小例，收尾落到把工具当 API 产品来做。
Agent 评测入门：怎么知道你的智能体到底行不行
Agent 的行为是多步、不确定、路径不唯一的，传统「对一次输出」的评测方法失效。本文给出评测分层框架、评测集构建方法、三种判分方式与 LLM-as-judge 的偏差缓解，最后把评测接进 CI。