diff --git a/cache/seen_urls.json b/cache/seen_urls.json
index 45e560f..00f78c3 100644
--- a/cache/seen_urls.json
+++ b/cache/seen_urls.json
@@ -1,5 +1,5 @@
 {
-  "lastUpdate": "2026-03-03T15:00:57.517Z",
+  "lastUpdate": "2026-03-04T15:01:07.708Z",
   "urls": [
     "http://arxiv.org/abs/2602.23360v1",
     "http://arxiv.org/abs/2602.23359v1",
@@ -273,6 +273,119 @@
     "https://github.com/login?return_to=%2fagentscope-ai%2fagentscope",
     "https://github.com/login?return_to=%2fagentscope-ai%2freme",
     "https://github.com/login?return_to=%2flmcache%2flmcache",
-    "https://github.com/login?return_to=%2faquasecurity%2ftrivy"
+    "https://github.com/login?return_to=%2faquasecurity%2ftrivy",
+    "http://arxiv.org/abs/2603.03280v1",
+    "http://arxiv.org/abs/2603.03278v1",
+    "http://arxiv.org/abs/2603.03258v1",
+    "http://arxiv.org/abs/2603.03252v1",
+    "http://arxiv.org/abs/2603.03242v1",
+    "http://arxiv.org/abs/2603.03241v1",
+    "http://arxiv.org/abs/2603.03233v1",
+    "http://arxiv.org/abs/2603.03230v1",
+    "http://arxiv.org/abs/2603.03224v1",
+    "http://arxiv.org/abs/2603.03212v1",
+    "http://arxiv.org/abs/2603.03206v1",
+    "http://arxiv.org/abs/2603.03203v1",
+    "http://arxiv.org/abs/2603.03195v1",
+    "http://arxiv.org/abs/2603.03190v1",
+    "http://arxiv.org/abs/2603.03180v1",
+    "http://arxiv.org/abs/2603.03177v1",
+    "http://arxiv.org/abs/2603.03176v1",
+    "http://arxiv.org/abs/2603.03175v1",
+    "http://arxiv.org/abs/2603.03163v1",
+    "http://arxiv.org/abs/2603.03158v1",
+    "http://arxiv.org/abs/2603.03155v1",
+    "http://arxiv.org/abs/2603.03147v1",
+    "http://arxiv.org/abs/2603.02123v2",
+    "http://arxiv.org/abs/2603.03146v1",
+    "http://arxiv.org/abs/2603.03143v1",
+    "http://arxiv.org/abs/2603.03142v1",
+    "http://arxiv.org/abs/2603.03140v1",
+    "http://arxiv.org/abs/2603.03131v1",
+    "http://arxiv.org/abs/2603.03119v1",
+    "http://arxiv.org/abs/2603.03116v1",
+    "http://arxiv.org/abs/2603.03112v1",
+    "http://arxiv.org/abs/2603.03106v1",
+    "http://arxiv.org/abs/2603.03101v1",
+    "http://arxiv.org/abs/2603.03099v1",
+    "http://arxiv.org/abs/2603.03249v1",
+    "http://arxiv.org/abs/2603.03205v1",
+    "http://arxiv.org/abs/2603.03202v1",
+    "http://arxiv.org/abs/2603.03198v1",
+    "http://arxiv.org/abs/2603.03194v1",
+    "http://arxiv.org/abs/2603.03192v1",
+    "http://arxiv.org/abs/2603.03134v1",
+    "http://arxiv.org/abs/2603.03111v1",
+    "http://arxiv.org/abs/2603.03096v1",
+    "http://arxiv.org/abs/2603.03095v1",
+    "http://arxiv.org/abs/2603.03081v1",
+    "http://arxiv.org/abs/2603.03072v1",
+    "http://arxiv.org/abs/2603.03056v1",
+    "http://arxiv.org/abs/2603.03054v1",
+    "http://arxiv.org/abs/2603.03047v1",
+    "http://arxiv.org/abs/2603.03001v1",
+    "http://arxiv.org/abs/2603.02983v1",
+    "http://arxiv.org/abs/2603.02945v1",
+    "http://arxiv.org/abs/2603.02909v1",
+    "http://arxiv.org/abs/2603.02876v1",
+    "http://arxiv.org/abs/2603.02873v1",
+    "http://arxiv.org/abs/2603.02865v1",
+    "http://arxiv.org/abs/2603.02860v1",
+    "http://arxiv.org/abs/2603.02842v1",
+    "http://arxiv.org/abs/2603.03281v1",
+    "http://arxiv.org/abs/2603.03275v1",
+    "http://arxiv.org/abs/2603.03270v1",
+    "http://arxiv.org/abs/2603.03269v1",
+    "http://arxiv.org/abs/2603.03259v1",
+    "http://arxiv.org/abs/2603.03251v1",
+    "http://arxiv.org/abs/2603.03238v1",
+    "http://arxiv.org/abs/2603.03235v1",
+    "http://arxiv.org/abs/2603.03234v1",
+    "http://arxiv.org/abs/2603.03229v1",
+    "http://arxiv.org/abs/2603.03227v1",
+    "http://arxiv.org/abs/2603.03226v1",
+    "http://arxiv.org/abs/2603.03211v1",
+    "http://arxiv.org/abs/2603.03207v1",
+    "http://arxiv.org/abs/2603.03196v1",
+    "http://arxiv.org/abs/2603.03191v1",
+    "http://arxiv.org/abs/2603.03188v1",
+    "http://arxiv.org/abs/2603.03172v1",
+    "http://arxiv.org/abs/2603.03135v1",
+    "https://huggingface.co/papers/2602.22600",
+    "https://huggingface.co/papers/2602.20672",
+    "https://huggingface.co/papers/2603.01059",
+    "https://huggingface.co/papers/2603.01421",
+    "https://huggingface.co/papers/2603.01697",
+    "https://huggingface.co/papers/2603.02765",
+    "https://huggingface.co/papers/2603.03281",
+    "https://huggingface.co/papers/2603.00618",
+    "https://huggingface.co/papers/2603.03163",
+    "https://huggingface.co/papers/2602.18671",
+    "https://huggingface.co/papers/2603.01655",
+    "https://huggingface.co/papers/2602.22752",
+    "https://huggingface.co/papers/2603.03205",
+    "https://huggingface.co/papers/2603.01683",
+    "https://huggingface.co/papers/2603.00656",
+    "https://huggingface.co/papers/2603.03195",
+    "https://huggingface.co/papers/2603.02252",
+    "https://huggingface.co/papers/2603.03283",
+    "https://huggingface.co/papers/2603.00729",
+    "https://huggingface.co/papers/2602.17100",
+    "https://huggingface.co/papers/2603.02398",
+    "https://huggingface.co/papers/2603.03194",
+    "https://huggingface.co/papers/2603.03202",
+    "https://huggingface.co/papers/2603.03142",
+    "https://huggingface.co/papers/2603.03276",
+    "https://huggingface.co/papers/2603.02573",
+    "https://huggingface.co/papers/2603.01400",
+    "https://huggingface.co/papers/2603.03241",
+    "https://huggingface.co/papers/2603.02175",
+    "https://huggingface.co/papers/2603.03160",
+    "https://github.com/login?return_to=%2fkeygraphhq%2fshannon",
+    "https://github.com/sponsors/msitarzewski",
+    "https://github.com/sponsors/fujiwarachoki",
+    "https://github.com/login?return_to=%2fitzcrazykns%2fperplexica",
+    "https://github.com/login?return_to=%2fnautechsystems%2fnautilus_trader",
+    "https://github.com/sponsors/flowiseai"
   ]
 }
\ No newline at end of file
diff --git a/daily/2026-03-04_en.md b/daily/2026-03-04_en.md
new file mode 100644
index 0000000..3b32bfb
--- /dev/null
+++ b/daily/2026-03-04_en.md
@@ -0,0 +1,80 @@
+# AI Daily Brief - 2026-03-04
+
+> Collected at: 3/4/2026, 11:01:07 PM
+> Total items: 113
+
+## 🔥 Top 10 Highlights
+
+1. AI-for-Science Low-code Platform with Bayesian Adversarial Multi-Agent Framework
+   http://arxiv.org/abs/2603.03233v1
+
+2. AgentConductor: Topology Evolution for Multi-Agent Competition-Level Code Generation
+   https://huggingface.co/papers/2602.17100
+
+3. Token Reduction via Local and Global Contexts Optimization for Efficient Video Large Language Models
+   https://huggingface.co/papers/2603.01400
+
+4. Speculative Speculative Decoding
+   http://arxiv.org/abs/2603.03251v1
+
+5. Conditioned Activation Transport for T2I Safety Steering
+   https://huggingface.co/papers/2603.03163
+
+6. Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use
+   https://huggingface.co/papers/2603.03205
+
+7. InfoPO: Information-Driven Policy Optimization for User-Centric Agents
+   https://huggingface.co/papers/2603.00656
+
+8. Qwen3-Coder-Next Technical Report
+   https://huggingface.co/papers/2603.00729
+
+9. BeyondSWE: Can Current Code Agent Survive Beyond Single-Repo Bug Fixing?
+   https://huggingface.co/papers/2603.03194
+
+10. Code2Math: Can Your Code Agent Effectively Evolve Math Problems Through Exploration?
+   https://huggingface.co/papers/2603.03202
+
+## 📂 Categories
+
+### AI Coding Tools / Code Agent
+
+- Code2Math: Can Your Code Agent Effectively Evolve Math Problems Through Exploration?
+  http://arxiv.org/abs/2603.03202v1
+- BeyondSWE: Can Current Code Agent Survive Beyond Single-Repo Bug Fixing?
+  http://arxiv.org/abs/2603.03194v1
+
+### Agent Frameworks
+
+- Saarthi for AGI: Towards Domain-Specific General Intelligence for Formal Verification
+  http://arxiv.org/abs/2603.03175v1
+- Learning When to Act or Refuse: Guarding Agentic Reasoning Models for Safe Multi-Step Tool Use
+  http://arxiv.org/abs/2603.03205v1
+- Learning to Generate and Extract: A Multi-Agent Collaboration Framework For Zero-shot Document-level Event Arguments Extraction
+  http://arxiv.org/abs/2603.02909v1
+
+### AI Infrastructure / Inference Optimization
+
+- Density-Guided Response Optimization: Community-Grounded Alignment via Implicit Acceptance Signals
+  http://arxiv.org/abs/2603.03242v1
+- SynthCharge: An Electric Vehicle Routing Instance Generator with Feasibility Screening to Enable Learning-Based Optimization and Benchmarking
+  http://arxiv.org/abs/2603.03230v1
+- Understanding and Mitigating Dataset Corruption in LLM Steering
+  http://arxiv.org/abs/2603.03206v1
+- Type-Aware Retrieval-Augmented Generation with Dependency Closure for Solver-Executable Industrial Optimization Modeling
+  http://arxiv.org/abs/2603.03180v1
+- Conditioned Activation Transport for T2I Safety Steering
+  http://arxiv.org/abs/2603.03163v1
+- Channel-Adaptive Edge AI: Maximizing Inference Throughput by Adapting Computational Complexity to Channel States
+  http://arxiv.org/abs/2603.03146v1
+- AI Space Physics: Constitutive boundary semantics for open AI institutions
+  http://arxiv.org/abs/2603.03119v1
+- MoD-DPO: Towards Mitigating Cross-modal Hallucinations in Omni LLMs using Modality Decoupled Preference Optimization
+  http://arxiv.org/abs/2603.03192v1
+- TAO-Attack: Toward Advanced Optimization-Based Jailbreak Attacks for Large Language Models
+  http://arxiv.org/abs/2603.03081v1
+- Adaptive Methods Are Preferable in High Privacy Settings: An SDE Perspective
+  http://arxiv.org/abs/2603.03226v1
+
+---
+*Generated by AINewsCollector*
diff --git a/daily/2026-03-04_zh.md b/daily/2026-03-04_zh.md
new file mode 100644
index 0000000..c4f4cfd
--- /dev/null
+++ b/daily/2026-03-04_zh.md
@@ -0,0 +1,80 @@
+# AI Daily Brief - 2026-03-04
+
+> 采集时间: 2026/3/4 23:00:18
+> 总条目: 113
+
+## 🔥 Top 10 重要消息
+
+1. 基于贝叶斯对抗多智能体框架的AI科学低代码平台
+   http://arxiv.org/abs/2603.03233v1
+
+2. 多智能体竞赛级代码生成：拓扑演进驱动AgentConductor
+   https://huggingface.co/papers/2602.17100
+
+3. 优化局部与全局上下文，高效精简视频大语言模型视觉标记
+   https://huggingface.co/papers/2603.01400
+
+4. 突破序列瓶颈：基于预测解码的加速推理技术
+   http://arxiv.org/abs/2603.03251v1
+
+5. 条件激活传输提升T2I安全引导
+   https://huggingface.co/papers/2603.03163
+
+6. 安全多步工具使用中，学习何时行动或拒绝：守护代理推理模型
+   https://huggingface.co/papers/2603.03205
+
+7. 信息驱动策略优化：面向用户中心的LLM智能体信息获取与决策
+   https://huggingface.co/papers/2603.00656
+
+8. Qwen3-Coder-Next：高效开源编程智能体语言模型
+   https://huggingface.co/papers/2603.00729
+
+9. 超越单库修复：代码智能体能否应对跨库推理挑战？
+   https://huggingface.co/papers/2603.03194
+
+10. Code2Math：代码智能生成奥赛数学难题探索新路径
+   https://huggingface.co/papers/2603.03202
+
+## 📂 分类汇总
+
+### AI 编程工具 / Code Agent
+
+- Code2Math：代码智能生成奥赛数学难题探索新路径
+  http://arxiv.org/abs/2603.03202v1
+- 超越单库修复：代码智能体能否应对跨库推理挑战？
+  http://arxiv.org/abs/2603.03194v1
+
+### Agent 框架
+
+- Saarthi：迈向领域特定通用智能的端到端形式化验证框架
+  http://arxiv.org/abs/2603.03175v1
+- 安全多步工具使用中，学习何时行动或拒绝：守护代理推理模型
+  http://arxiv.org/abs/2603.03205v1
+- 多智能体协作框架助力零样本文档级事件论元提取
+  http://arxiv.org/abs/2603.02909v1
+
+### AI 基础设施 / 推理优化
+
+- 基于密度引导的响应优化：通过隐式接受信号实现社区基础对齐
+  http://arxiv.org/abs/2603.03242v1
+- “SynthCharge：基于可行性筛选的电动汽车路径实例生成器，助力学习优化与基准测试”
+  http://arxiv.org/abs/2603.03230v1
+- LLM引导中数据集腐败理解与缓解：对比引导法提升生成行为
+  http://arxiv.org/abs/2603.03206v1
+- 基于依赖闭包的求解器可执行工业优化建模的Type-Aware检索增强生成
+  http://arxiv.org/abs/2603.03180v1
+- 条件激活传输提升T2I安全引导
+  http://arxiv.org/abs/2603.03163v1
+- 自适应信道边缘AI：通过适应信道状态最大化推理吞吐量
+  http://arxiv.org/abs/2603.03146v1
+- 开放AI机构构建：基于构成边界语义的代理AI部署
+  http://arxiv.org/abs/2603.03119v1
+- MoD-DPO：基于模态解耦偏好优化的全模态LLMs跨模态幻觉缓解
+  http://arxiv.org/abs/2603.03192v1
+- TAO-Attack：针对大型语言模型的先进优化型越狱攻击研究
+  http://arxiv.org/abs/2603.03081v1
+- 高隐私环境下，自适应优化方法优于传统方法：随机微分方程视角
+  http://arxiv.org/abs/2603.03226v1
+
+---
+*Generated by AINewsCollector*