1
The LLM rewrote the email after I approved it
AI代理篡改已批准邮件？600万小时运行揭示真正痛点不在模型智力，而在边界失控。
Article URL: https://www.crusbro.com/en/product-architecture.html Comments URL: https://news.ycombinator.com/item?id=49384800 Points: 2 # Comments: 0
2
Why Most Multi-Agent Systems Fail in Production (And How to Fix It)
多智能体系统在测试环境跑得通，生产环境却频繁崩溃？根因不在模型，而在编排层——实战经验教你避开无限重试与静默故障的坑。
Most multi-agent demos look impressive on stage. Then they hit production and fall apart. Here's the pattern: agents that "worked" in a Jupyter notebo…
3
Macroeconomic Forecasting with Large Language Models
大模型也能预测GDP和通胀？这项研究把LLM用在了宏观经济预测上，看它如何从海量文本中捕捉经济动向。
arXiv:2407.00890v5 Announce Type: replace-cross Abstract: This paper presents a comparative analysis evaluating the accuracy of Large Language Models …
4
Examples for use of AI and especially LLMs in major mathematical developments
盘点大模型参与重大数学突破的真实案例，看AI如何助力数学研究前沿。
Article URL: https://mathoverflow.net/questions/502120/examples-for-the-use-of-ai-and-especially-llms-in-major-mathematical-development Comments URL: …
5
Show HN: Anansi – open-source memory API for LLM apps
开源记忆API，为LLM应用提供持久上下文，五分钟自托管部署，让AI记住该记住的。
Article URL: https://github.com/g-33-L/anansi Comments URL: https://news.ycombinator.com/item?id=49256253 Points: 1 # Comments: 0
6
Towards an LLM-based method for quantifying the sexual content in song lyrics
用大模型量化雷鬼歌曲歌词的性内容，方法可复现，为音乐文本研究提供新工具
arXiv:2608.08885v1 Announce Type: cross Abstract: Reggaeton is one of the most widely consumed music genres in the world, and its lyrics are commonly …
7
Ask HN: Do you use LLMs for non-coding related work?
探讨LLM在编程之外的真实工作场景，看开发者们如何用AI处理日常任务，实用经验与吐槽并存
LLMs are very good at code. We can argue on the merits of their other qualities, but we can all agree on that. However, code is a very small section o…
8
AiFlow: Token-Native Reactive Orchestration with Bounded Backpressure for Streaming LLM Applications
面向流式LLM应用的Token原生编排方案，用有界背压解决生成式推理的流量控制难题，架构思路值得开发者借鉴。
arXiv:2608.00558v1 Announce Type: cross Abstract: Large language model (LLM) applications increasingly operate as streaming workflows combining retrie…
9
Why Silicon Valley's 'Neurodiversity Hiring' Programs Are Failing Autistic Workers (And How LLMs Are Quietly Fixing the Interview)
硅谷神经多样性招聘为何失灵？LLM正在悄然改写面试规则，让自闭症人才不再被埋没。
Why Silicon Valley's 'Neurodiversity Hiring' Programs Are Failing Autistic Workers (And How LLMs Are Quietly Fixing the Interview) Introduction: The G…
10
Show HN: Grounded-forge – RAG with summaries and task views precomputed ingest
结构化9-pass预计算，让RAG运行时直接查索引而非重新推导，效率拉满的开源新方案。
Hi everyone, I've been working on a knowledge retrieval system that can create distributable applications for several months and I'd love to share it …
11
Idea2Plan: Exploring AI-Powered Research Planning
AI大模型从想法到研究计划，探索自动化科研规划新范式
arXiv:2510.24891v2 Announce Type: replace-cross Abstract: Large language models (LLMs) have demonstrated significant potential to accelerate scientifi…
12
Building Reliable LLM Applications in Java
用 Java 结合 Claude 构建生产级 LLM 应用，教你直接获取结构化 JSON 输出，告别手动解析
Introduction LLMs are usually associated with Python, but a great deal of production software — banking, enterprise backends, long-lived services — ru…
13
Building Reliable LLM Applications in Python
用Python打造可预测、无幻觉的LLM应用工程实践，揭秘Anthropic Claude生产级部署要点
Introduction Calling an LLM API is easy. Building an application on top of one that is reliable — that fails predictably, doesn't hallucinate its way …
14
I Compared the Real Cost of Claude Code, OpenRouter, and Image APIs
手把手算清三大AI服务的月费账单，开发者省钱必看。
An API request that looks cheap on a pricing page can become much more expensive inside a real product. The pricing page normally gives you the unit r…
15
A Lightweight Multi-Agent Framework for Automated Concrete Barrier Design
多智能体框架自动化符合AASHTO规范的混凝土护栏设计，终结繁琐人工迭代。
arXiv:2606.12040v1 Announce Type: new Abstract: The design of reinforced concrete highway barriers is a safety-critical process that requires strict c…
16
Evaluating Agentic Configuration Repair for Computer Networks
首个系统性评估AI代理自动修复计算机网络配置错误的能力，揭示代理性能与关键影响因素
arXiv:2606.06212v1 Announce Type: new Abstract: Misconfigurations in computer networks remain a major source of critical Internet outages. Research is…
17
Parthenon Law: A Self-Evolving Legal-Agent Framework
法律AI代理新框架，自我进化解决端到端法律事务的三大障碍。
arXiv:2606.04602v1 Announce Type: new Abstract: As agents grow more capable, legal-domain LLM agents promise to turn document-heavy matters into revie…
18
PropLLM: Propagation-Aware Scene Reconstruction for Network Fault Diagnosis
融合传播感知与LLM的场景重建方法，专为网络故障诊断设计，有望提升根因定位效率。
arXiv:2606.00582v1 Announce Type: new Abstract: Network faults propagate layer by layer along topology and protocol dependencies, yet operations syste…
19
Headroom compresses everything your AI agent reads before it reaches the LLM
AI代理读取内容前自动压缩，最高减少90% token消耗，实现成本断崖式下降。
Article URL: https://pypi.org/project/headroom-ai/ Comments URL: https://news.ycombinator.com/item?id=48349275 Points: 1 # Comments: 0
20
Toxicity in Twitch Chats: An LLM-Based Analysis Across Gaming Communities
用大模型分析Twitch各游戏社区聊天毒性差异，揭示不同社区的独特模式。
arXiv:2605.24000v1 Announce Type: new Abstract: Toxicity in online gaming communities remains a persistent challenge, manifesting across genres, platf…