whshang

Uninstalled Orca today and moved fully to the new Herdr desktop app. Orca served me well and the team ships fast, so no complaints there. What tipped it for me was how each tool tracks agents. Orca puts hooks into every agent's config. When I cleaned up, I found 13 in Claude Code, 9 in Codex, 8 in Devin and a plugin in Hermes, all running on every prompt and tool call. Herdr reads agent status from the terminal itself. My Claude and Codex configs are clean again: no hooks, nothing running between me and the agent. What I like most: • Agents can drive the workspace through a small CLI. Claude opens a pane next to itself and leaves a command at my prompt (sudo installs, prod releases) without running it. I just press Enter. • Spaces, tabs and panes per project, with Claude and Codex side by side • Less setup and fewer moving parts I still use the Claude and Codex desktop apps too, but for terminal agent work Herdr is home now.

在陪伴孩子成长的过程中，我想记住重要的瞬间，也想看清自己在亲子沟通上做得不够的地方。我和伙伴们一起做了 Pastok ，帮我记住日常中的语音沟通，助我日省吾身。当前支持 Apple Watch 和 iPhone。有类似需求的朋友不妨试试，欢迎反馈意见。 Let the past talk. pastok.com

过去一个月是疯狂的一个月 大概一个月前，我下定决心重构 kimi-code，开始设计新的架构。 我大概抱着电脑和便携屏在汤泉卷了两整天，花了几千刀的 token 去做架构分析、设计和验证，最终得到了一份我认为最优的架构方案。 我觉得在 vibe 时代，架构变得更加重要了，一份好的架构能够在可控的范围内，让 Agent 肆意 coding，而不会打破东西 - 架构确定后，就开始冲刺实现。（过程中吵和推翻了无数次） - 迅速组建了一个强大的 team，感恩兄弟们无条件的信任🙇‍♂️ - 迅速 onboarding 整个 team，🙇‍♂️ 再次感恩兄弟们 - 封闭开发了一段时间（🤣年轻的时候，觉得是糟粕，真到时候，发现是人类工程效率奇迹。你无法想象随时可以拉着全部人在白板前吵架的架构迭代速度） - 虽然代码都是 vibe 的，但依旧逃不过 “代码质量正比于人类的注意力密度”。所以 agent 并不会替代所有程序员，只会让顶级的程序员生产力翻 20 倍，并淘汰其他程序员，且，集体主义 >>> 个人英雄主义。 - 一步一个坑的解决过程中遇到的问题。每一天都是最绝望的一天😭 - 开源后就病倒了，皮质醇分泌过度，影响免疫力 - 这一个月学的东西够我消化半年的 - 一周干了一整箱红牛，还得是生物燃料 - 🫥 也在 x 上消失了一个月 本来想写一些文章去总结过程中一些 insights 和 idea，但我本来就不擅长写长文，外加人脑自我保护让我迅速忘记了整个过程中的痛苦，并模糊了时间观念（冷知识，kimi-code 重构版开源其实才过了一周多，但在我的感性认知中，像是已经过了一个月） 等 kimi-code 陆续迭代到稳定，再去总结过程中的 lessons learned

🚀 Better inference efficiency, lower costs, broader access. MiMo-V2.5 Series API pricing is now permanently reduced — by up to 99% compared to previous pricing. ✨ Unified pricing across all context lengths. MiMo Token Plans have also been upgraded: • 5–8× more usable tokens at the same price • Simpler and more transparent billing rules 🎁 As a thank-you to current users, all current Token Plan credits will be fully reset. 🎧 MiMo-V2.5-TTS remains free for a limited time. ⏰ Effective May 26 at 6:00 PM PDT. These improvements are powered by continued inference optimization and serving efficiency upgrades across the MiMo stack. 🛠️ We’ll also publish a detailed technical blog on the inference optimizations later — stay tuned.

OpenCLI 一周生日了，4.2K star，40+ contributor，做个总结吧。github.com/jackwener/opencli 目前已经支持 OpenCLI 通过 CLI操作 50+ 网站，twitter，reddit，bilibili，Bloomberg，LinkedIn....社交、新闻、金融，都覆盖了。 譬如可以通过 opencli twitter bookmarks 一键获取推特收藏，可以让 AI agent 调用它进行总结，分析等等 支持通过 CLI 直接操作 electron 应用进行操作。 譬如可以 opencli notion search xxx 进行搜索 CLI hub 能力，通过 opencli 集成所有的 CLI 充当他们的说明书，AI Agent 执行一句 opencli list 就能发现你所有的工具，Agent 立刻学会调用。 譬如可以 opencli gh status 一行命令下载 推特图片、B站视频、小红书笔记、知乎文章 —— 全都能下载，配合 openclaw 等非常方便采集信息。 最近 @browser_use 很火，很多人问我有什么区别 - 输入输出的确定性 -- 输出结构固定，错误清晰可复现 - CLI 本身不消耗 token，探索一次网站后就可以沉淀 CLI 反复使用 - CLI hub 能力，接入所有 CLI，成为 AI agent 的 CLI 说明书 - 丰富的生态，50平台适配器，大量丰富的命令 - electron 桌面应用控制 -- CDP + AppleScript 控制应用

GLM-5 from @Zai_org is now available in TRAE. Users can now integrate GLM-5 as a custom model through Z.ai provider route. Settings -> Models -> Add Model

Codex 5.3 vs Opus 4.6 tldr; opus 4.6: + 1M context + enterprise/knowledge work + 500 zero-days found + agent teams in claude code - not benching as high as codex 5.3 **idgaf about self reported benches gpt-5.3-codex + wins code benchmarks + faster + mid-task steering !! - less than half the context window of opus gonna have to try codex ig

🚀 Introducing Qwen3-Coder-Next, an open-weight LM built for coding agents & local development. What’s new: 🤖 Scaling agentic training: 800K verifiable tasks + executable envs 📈 Efficiency–Performance Tradeoff: achieves strong results on SWE-Bench Pro with 80B total params and 3B active ✨ Supports OpenClaw, Qwen Code, Claude Code, web dev, browser use, Cline, etc 🤗 Hugging Face: huggingface.co/collections/Q… 🤖 ModelScope: modelscope.cn/collections/Qw… 📝 Blog: qwen.ai/blog?id=qwen3-coder-… 📄 Tech report: github.com/QwenLM/Qwen3-Code…

To celebrate the Codex app, we're launching a promo: - Doubled Codex rate limits for all paid plans (2 months) - Access for ChatGPT Free and Go Plans (1 month) Check it out at openai.com/codex

MCP servers can bloat your context window, so we built a solution: the MCP search subagent. This tool reduces context usage by up to 26% in our testing, and works with *any* AI model. And we had had product engineer Matthew Albright explain how we built it (in thread 🧵)

GLM-4.7 is here! GLM-4.7 surpasses GLM-4.6 with substantial improvements in coding, complex reasoning, and tool usage, setting new open-source SOTA standards. It also boosts performance in chat, creative writing, and role-play scenarios. Default Model for Coding Plan: z.ai/subscribe Try it now: chat.z.ai Weights: huggingface.co/zai-org/GLM-4… Tech Blog: z.ai/blog/glm-4.7

You can now use Claude Code with OpenRouter 🎊 Code with over 320 LLMs, including 39 free ones!

Limited-Time Offer: KAT-Coder-Pro V1 Free Trial Since the launch of the KAT-Coder model, we have received extensive positive feedback from our users. This has been instrumental in driving the rapid iteration and optimization of our model. To show our appreciation for your support and to provide everyone with an opportunity to fully experience the product, we are offering complimentary access to the KAT-Coder-Pro V1 model, effective immediately. This free access will be available until 11:00:00 on November 10 (EST, UTC-5)/00:00:00 on November 11(Beijing Time, UTC+8).🥳 We sincerely encourage you to use the model extensively and continue providing us with your valuable feedback. Let's work together to drive the advancement and optimization of AI coding models.