alazycoder

今天 GitHub 被被 Agent 军团彻底屠榜了🚀 5 个星标暴增最狠的项目，专业拆解下！ 1. MervinPraison/PraisonAI 
低代码多代理框架，把 LLM 直接变成你的 24/7 AI员工团队，支持规划、研究、编码、交付，还能跨 Telegram/Discord/WhatsApp 无缝交接、带记忆RAG 和 100+模型。 
🔗 直达 github.com/MervinPraison/… 
昨天还在手动敲代码，今天直接组了个AI打工人军团、程序员们看着它自己跑任务只能原地社死！新增1842星，军团已就位！ 
🟢 2. vectorize-io/hindsight 
Agent 专属学习型记忆引擎，彻底解决长任务忘事难题，让自主Agent真正拥有持久上下文和自我进化能力。 
🔗 直达 github.com/vectorize-io/h… 
以前 Agent 跑两小时就失忆，现在直接开终身学习模式，程序员终于不用半夜爬起来喂 prompt 了。新增 1237星，这记忆力比我前女友都强！ 
🟢🟢 3. hsliuping/TradingAgents-CN 
基于多智能体 LLM 的中文金融交易框架，多 Agent 协作实时决策、风控、执行，专治一个人盯盘盯到吐的痛点。 
🔗 直达 github.com/hsliuping/Trad… 
以前交易员 996 盯盘，现在多 Agent 自己组队操盘，程序员看一眼就想辞职去开量化私募。新增 987星，金融圈要集体失业了！ 
🟢🟢🟢 4. microsoft/apm 
Agent Package Manager，一键管理、安装、编排所有 AI Agent工具包，像npm 一样玩转 Agent 生态 
🔗 直达 github.com/microsoft/apm 
以前搭个 Agent 环境要手动装半天，现在直接 apm install agent-army，程序员爽到飞起。新增 756星，微软这是要让 Agent 像 npm 一样无处不在啊！ 
🟢🟢🟢🟢 5. Alishahryar1/free-claude-code 
完全免费开源的 Claude Code 替代，在终端/VSCode/Discord 直接跑顶级编码Agent，像 OpenClaw 一样但零成本 
🔗 直达 github.com/Alishahryar1/f…
 Anthropic 刚想收紧，社区直接把编码Agent全开源了，程序员们狂呼免费午餐来了。新增643星，付费墙直接被炸穿！ ⚠️⚠️ 总结 这五个项目不是零散工具，而是Agent军团的完整闭环。记忆、包装、协作、实战、编码全覆盖，2026年、一个人顶一个团队不再是梦，直接把你的AI打工人军团拉满🚀🦞

pwru is an eBPF-based tool for tracing network packets in the Linux kernel with advanced filtering capabilities It allows fine-grained introspection of kernel state to facilitate debugging network connectivity issues ➤ ku.bz/Q3X1ngZGC

k8sql is a tool written in Rust that queries Kubernetes clusters using SQL syntax, exposing resources as database tables and kubectl contexts as databases with a query planner that optimizes API calls by pushing filters server-side ➤ ku.bz/fn8FJqySQ

Cling 2.0 - 在50毫秒内模糊搜索任意文件 官方说能在 900 万+ 文件里做到 <100ms 级搜索，很猛。 github.com/FuzzyIdeas/Cli…

Yet another incredible RSpec to Minitest success story

在 GitHub 上看到一份专门针对 Agent 技能编写的最佳实践指南：skills-best-practices。 总结了一套标准化的目录结构，把核心指令、执行脚本和参考文件梳理得清清楚楚。 详细讲解如何优化元数据描述，巧妙利用负面触发词，大幅提高技能被精准调用的概率。 GitHub：github.com/mgechev/skills… 还提供了一套实用的验证提示词，教我们如何直接让大模型来测试技能的逻辑闭环和边缘场景。 核心思路是把复杂的重复操作交给独立脚本，让大模型只做高层决策，有效控制 token 消耗。 内容质量高且落地强，想要撰写 Skill 的朋友，这份指南可以收藏学习一下。

Ruby 3.2.11 has been released ruby-lang.org/en/news/2026/0… This release is the final version of the 3.2 series. From Christmas 2022 to today, three years of contributions made Ruby better. Thank you all.

Did you know you can freeze and resume processes on Linux? 🧙‍♂️ Freezing a process lets you safely pause it without stopping or killing it. This can be useful if you want to temporarily suspend a heavy job during peak load, hold a service still while you reconfigure its environment, or "stop the world" to inspect the system's state during a security audit or a bug investigation. Works for both processes and containers! Try it out - using the cgroup freezer is surprisingly easy: - Freeze and thaw a Linux process labs.iximiuz.com/challenges/lin… - Pause and unpause a containerized app labs.iximiuz.com/challenges/doc…

Kthena is a Kubernetes-native platform for deploying and managing Large Language Models with support for multiple inference engines like vLLM, SGLang, and Triton ➤ ku.bz/HL_6Y0kNj

想深入了解 Harness 工程实践，推荐关注 yoyo 的实现。 我们不但开源了所有代码，也全程开源了 yoyo 的进化史。你可以像看纪录片一样，见证他在 26 天里，从一个仅 200 行代码的‘雏形代理’，通过 递归自我进化（Self-Evolution）与自主编码，成长为拥有 34706 行代码、性能直逼 Claude Code 的全能型 Agent。 yoyo 的进化不只是代码量的堆砌，而是一场目标驱动的野蛮生长。如何平衡这种生长的‘野性’与工程的‘稳定性’，是我们面临的最大课题。此外，关于其人格特质（Personality）的塑造、多级记忆架构（L/M/S Term Memory）以及社交属性的演进，都极具技术探讨价值。这一路攒下的‘坑’和经验，我打算写篇长文和大家好好聊聊，深度分享背后的工程思考。 Repo: github.com/yologdev/yoyo-…

AI is bad with Hotwire. @julian_rubisch is trying to change it. For years he was building a library of advanced Hotwire behaviors hotwire.club Now, it's open source! Add these skills to your app and turn your Hotwire from 💩 to 🥇 Thanks Julian! 📷🙏

只有像我们这种老程序员才懂得这个视频的含金量啊

rpg 0.8.0 – modern PostgreSQL terminal github.com/NikolayS/rpg in this release: EXPLAIN plans that actually help you Instead of staring at a wall of text, rpg now shows a summary header with execution time, buffer stats, and automatic warnings — seq scans on large tables, row estimate errors, sorts spilling to disk. The full plan is preserved with color coding so the hot path jumps out instantly. Plan export to popular visualization/analysis tools When you find a slow query, share the plan in one command: \explain share depesz → explain.depesz.com/s/abc123 Works with depesz, dalibo, and pgMustard. No copy-paste, no browser tabs. Lua custom commands Write a Lua script, drop it in ~/.config/rpg/commands/, and you have a new \backslash command. Other highlights - Markdown output (\pset format markdown; psql --markdown) — paste query results directly into GitHub issues - Better \s history — filter with patterns, syntax highlighted, through the pager - Tab completion for all rpg commands - pspg integration with 20+ themes - lots of fixes based on feedback I've got yesterday/today ⭐ github.com/NikolayS/rpg

Rails default Dockerfile got faster and you might want to apply this change to your current projects github.com/rails/rails/pu…

本地OCR现在已经足够强大，不再需要依赖云端。GLM-OCR在Mac Studio M2 Ultra上，只需约2GB显存，即可处理表格、数学方程，每秒达约260个token。 这意味着，拥有GPU或Mac的你已经准备好迎接AI时代的到来。随着本地模型的快速升级，技术门槛正在迅速降低，未来将不再局限于大公司和专业人士。