<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>AI Daily Digest</title><description>每日精选 AI 工具动态 · 自动聚合自 6 大平台</description><link>https://ai-daily-digest.pages.dev/</link><item><title>Show HN: Uruky (EU-based Kagi alternative) now has Image Search and URL Rewrites</title><link>https://uruky.com/?il=en</link><guid isPermaLink="true">https://uruky.com/?il=en</guid><description>Uruky 是一款总部位于欧盟的新型搜索引擎，作为 Kagi 的替代品，新上线了图像搜索和 URL 重写功能，并提供通过工作量证明验证码获取的 2 小时免费试用。其源代码采用源可用（source-available）许可模式，已放弃强制签署 NDA/NCC 的要求。</description><pubDate>Thu, 04 Jun 2026 08:56:10 GMT</pubDate><source>hackernews</source><score>8.8</score></item><item><title>RepoRecon: I Got Tired of Reading Bad Code, So I Built an AI That Does It For Me</title><link>https://dev.to/az-dev1/reporecon-i-got-tired-of-reading-bad-code-so-i-built-an-ai-that-does-it-for-me-5djp</link><guid isPermaLink="true">https://dev.to/az-dev1/reporecon-i-got-tired-of-reading-bad-code-so-i-built-an-ai-that-does-it-for-me-5djp</guid><description>RepoRecon 是一款面向开发者的AI工具，旨在自动分析和理解代码仓库中的低质量或难以阅读的代码，减轻人工审查负担。</description><pubDate>Thu, 04 Jun 2026 06:14:22 GMT</pubDate><source>devto</source><score>9.2</score></item><item><title>I built a vulnerable app and spent $1,500 seeing if LLMs could hack it</title><link>https://kasra.blog/blog/i-spent-1500-seeing-if-llms-could-hack-my-app/</link><guid isPermaLink="true">https://kasra.blog/blog/i-spent-1500-seeing-if-llms-could-hack-my-app/</guid><description>作者构建了一个存在安全漏洞的Web应用，并投入1500美元测试主流大语言模型（LLM）是否能自主发现并利用这些漏洞，结果揭示了当前LLM在真实渗透测试任务中的能力边界与局限性。</description><pubDate>Thu, 04 Jun 2026 00:56:32 GMT</pubDate><source>hackernews</source><score>9.5</score></item><item><title>I Built an AI-Powered Meeting Platform From Scratch — Here’s How It Actually Works</title><link>https://dev.to/anupam_kumar/i-built-an-ai-powered-meeting-platform-from-scratch-heres-how-it-actually-works-31p</link><guid isPermaLink="true">https://dev.to/anupam_kumar/i-built-an-ai-powered-meeting-platform-from-scratch-heres-how-it-actually-works-31p</guid><description>本文详细介绍了从零构建AI驱动会议平台Hoovik的技术实现，涵盖WebRTC信令、基于Redis的分布式Node.js架构及实时情绪分析等关键模块。</description><pubDate>Wed, 03 Jun 2026 19:33:42 GMT</pubDate><source>devto</source><score>9.9</score></item><item><title>Gemma 4 12B: A unified, encoder-free multimodal model</title><link>https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12b/</link><guid isPermaLink="true">https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemma-4-12b/</guid><description>本文介绍Gemma 4 12B——一款统一的、无需独立编码器的多模态大模型，强调其架构创新与跨模态能力。该模型尚未开源或商用，目前仅以技术概念形式在社区披露。</description><pubDate>Wed, 03 Jun 2026 16:04:42 GMT</pubDate><source>hackernews</source><score>10</score></item><item><title>Every tool seems to have a coding agent horned in these days..... I don&apos;t think that makes sense.</title><link>https://dev.to/ben/every-tool-seems-to-have-a-coding-agent-horned-in-these-days-i-dont-think-that-makes-sense-3db</link><guid isPermaLink="true">https://dev.to/ben/every-tool-seems-to-have-a-coding-agent-horned-in-these-days-i-dont-think-that-makes-sense-3db</guid><description>本文是一篇对当前AI工具过度集成编码代理功能现象的批判性评论，质疑其实际价值与用户体验合理性。</description><pubDate>Wed, 03 Jun 2026 13:46:51 GMT</pubDate><source>devto</source><score>8.5</score></item><item><title>Run AI Coding Agents Safely with Docker Sandboxes</title><link>https://dev.to/pradumnasaraf/run-ai-coding-agents-safely-with-docker-sandboxes-81g</link><guid isPermaLink="true">https://dev.to/pradumnasaraf/run-ai-coding-agents-safely-with-docker-sandboxes-81g</guid><description>本文介绍了如何使用 Docker 容器沙箱安全地运行 AI 编程代理，防止其对宿主系统造成未授权访问或破坏，涵盖配置、权限限制和网络隔离等实操要点。</description><pubDate>Wed, 03 Jun 2026 13:15:02 GMT</pubDate><source>devto</source><score>9.5</score></item><item><title>Uber&apos;s $1,500/month AI limit is a useful signal for AI tool pricing</title><link>https://simonwillison.net/2026/Jun/3/uber-caps-usage/</link><guid isPermaLink="true">https://simonwillison.net/2026/Jun/3/uber-caps-usage/</guid><description>本文讨论优步对AI工具（如Claude Code）设置每月1500美元使用上限的商业决策，将其视为AI企业服务定价趋势的重要信号，反映大公司对AI成本管控的现实考量。</description><pubDate>Wed, 03 Jun 2026 12:25:14 GMT</pubDate><source>hackernews</source><score>9</score></item><item><title>The Query Was Still a Lie. The Tool Call Told the Truth.</title><link>https://dev.to/zep1997/the-query-was-still-a-lie-the-tool-call-told-the-truth-ahb</link><guid isPermaLink="true">https://dev.to/zep1997/the-query-was-still-a-lie-the-tool-call-told-the-truth-ahb</guid><description>本文探讨AI智能体系统中“查询失真”与“工具调用真实”之间的张力，指出仅依赖相关性评估不足以保障推理可靠性，并延续了自校正系统（Self-Correcting Systems）的前沿研究脉络。</description><pubDate>Wed, 03 Jun 2026 11:51:22 GMT</pubDate><source>devto</source><score>10.8</score></item><item><title>Your Agent Failed in Prod. Good Luck Reproducing It.</title><link>https://dev.to/tisha_chawla/your-agent-failed-in-prod-good-luck-reproducing-it-56ci</link><guid isPermaLink="true">https://dev.to/tisha_chawla/your-agent-failed-in-prod-good-luck-reproducing-it-56ci</guid><description>本文探讨了LLM智能体在生产环境中难以复现的根本原因，指出部分不确定性实为设计特性，并介绍了“记录与回放”（record-and-replay）这一实用调试方法，帮助开发者高效定位和修复问题。</description><pubDate>Wed, 03 Jun 2026 10:16:43 GMT</pubDate><source>devto</source><score>10.8</score></item><item><title>Claude 4 New Features: Complete Guide</title><link>https://www.youtube.com/watch?v=example1</link><guid isPermaLink="true">https://www.youtube.com/watch?v=example1</guid><description>Claude 4 发布全新功能，包括增强的代码生成能力和多模态推理，本视频详细演示了实际使用方法。</description><pubDate>Wed, 03 Jun 2026 06:00:00 GMT</pubDate><source>youtube</source><score>11.8</score></item><item><title>I built an AI agent that automates my entire workflow</title><link>https://reddit.com/r/LocalLLaMA/comments/example</link><guid isPermaLink="true">https://reddit.com/r/LocalLLaMA/comments/example</guid><description>开发者分享了使用本地 LLM 构建自动化工作流的完整方案，包含具体的模型选择和部署步骤。</description><pubDate>Wed, 03 Jun 2026 04:00:00 GMT</pubDate><source>reddit</source><score>10.2</score></item><item><title>Do Text Edits Generalize to Visual Generation? Benchmarking Cross-Modal Knowledge Editing in UMMs</title><link>https://huggingface.co/papers/2606.00477</link><guid isPermaLink="true">https://huggingface.co/papers/2606.00477</guid><description>本文提出UniKE基准，首次系统评估统一多模态模型（UMMs）中跨模态知识编辑的有效性，探究文本知识编辑能否泛化至图像生成任务。</description><pubDate>Fri, 29 May 2026 20:00:00 GMT</pubDate><source>huggingface</source><score>8.1</score></item><item><title>NousResearch/hermes-agent</title><link>https://github.com/NousResearch/hermes-agent</link><guid isPermaLink="true">https://github.com/NousResearch/hermes-agent</guid><description>NousResearch/hermes-agent 是一个开源的、可扩展的智能体框架，旨在随用户需求演进，支持动态工具调用与上下文自适应。该项目提供代码仓库、配置示例和基础文档，便于开发者快速集成与定制。</description><pubDate>Tue, 22 Jul 2025 22:22:28 GMT</pubDate><source>github</source><score>11.2</score></item><item><title>open-webui/open-webui</title><link>https://github.com/open-webui/open-webui</link><guid isPermaLink="true">https://github.com/open-webui/open-webui</guid><description>Open WebUI 是一个开源的、用户友好的本地化大模型交互界面，支持 Ollama、OpenAI API 等多种后端，便于快速部署和使用私有化 AI 服务。</description><pubDate>Fri, 06 Oct 2023 22:08:27 GMT</pubDate><source>github</source><score>8.6</score></item><item><title>ollama/ollama</title><link>https://github.com/ollama/ollama</link><guid isPermaLink="true">https://github.com/ollama/ollama</guid><description>Ollama 是一个开源的本地大模型运行框架，支持一键拉取和运行包括 Kimi-K2.6、GLM-5.1、Qwen、Gemma 等多个主流开源与商业对齐模型。</description><pubDate>Mon, 26 Jun 2023 19:39:32 GMT</pubDate><source>github</source><score>9.5</score></item><item><title>microsoft/generative-ai-for-beginners</title><link>https://github.com/microsoft/generative-ai-for-beginners</link><guid isPermaLink="true">https://github.com/microsoft/generative-ai-for-beginners</guid><description>这是微软在 GitHub 上发布的面向初学者的生成式 AI 实践教程，包含 21 节结构化课程，涵盖基础概念、API 调用、提示工程及小型应用开发。</description><pubDate>Mon, 19 Jun 2023 16:28:59 GMT</pubDate><source>github</source><score>12.1</score></item><item><title>langgenius/dify</title><link>https://github.com/langgenius/dify</link><guid isPermaLink="true">https://github.com/langgenius/dify</guid><description>Dify 是一个面向生产环境的开源平台，用于构建和部署基于智能体（agentic）的工作流，支持可视化编排、模型集成与应用发布。</description><pubDate>Wed, 12 Apr 2023 07:40:24 GMT</pubDate><source>github</source><score>9.5</score></item><item><title>Significant-Gravitas/AutoGPT</title><link>https://github.com/Significant-Gravitas/AutoGPT</link><guid isPermaLink="true">https://github.com/Significant-Gravitas/AutoGPT</guid><description>AutoGPT 是一个开源的自主AI代理框架，旨在降低AI应用与开发门槛，支持用户基于LLM构建可自主规划、执行和迭代的任务系统。</description><pubDate>Thu, 16 Mar 2023 09:21:07 GMT</pubDate><source>github</source><score>11.3</score></item><item><title>f/prompts.chat</title><link>https://github.com/f/prompts.chat</link><guid isPermaLink="true">https://github.com/f/prompts.chat</guid><description>f/prompts.chat 是一个开源的社区驱动型提示词（prompt）共享平台，前身是 Awesome ChatGPT Prompts，支持自托管以保障数据隐私。</description><pubDate>Mon, 05 Dec 2022 13:54:13 GMT</pubDate><source>github</source><score>10</score></item><item><title>langchain-ai/langchain</title><link>https://github.com/langchain-ai/langchain</link><guid isPermaLink="true">https://github.com/langchain-ai/langchain</guid><description>LangChain 是一个用于构建基于大语言模型的应用程序的开源框架，专注于代理（agent）工程、链式调用和工具集成。</description><pubDate>Mon, 17 Oct 2022 02:58:36 GMT</pubDate><source>github</source><score>9.4</score></item><item><title>huggingface/transformers</title><link>https://github.com/huggingface/transformers</link><guid isPermaLink="true">https://github.com/huggingface/transformers</guid><description>Hugging Face Transformers 是一个广泛使用的开源库，提供数千种预训练模型和统一API，支持文本、视觉、音频及多模态任务的推理与训练。</description><pubDate>Mon, 29 Oct 2018 13:56:00 GMT</pubDate><source>github</source><score>10.5</score></item></channel></rss>