熊貓隨口說

GitHub - Nunchi-trade/agent-cli

Contribute to Nunchi-trade/agent-cli development by creating an account on GitHub.

True

https://fixupx.com/pmarca/status/2030949039345467570?s=46&t=1LAyoawP6LK1AbrwCvLGqQ

9 小時前

FixupX

Marc Andreessen 🇺🇸 (@pmarca)

My information consumption is now 1/4 X, 1/4 podcast interviews of the smartest practitioners, 1/4 talking to the leading AI models, and 1/4 reading old books. The opportunity cost of anything else is far too high, and rising daily.

10 小時前

https://fixupx.com/realmcore_/status/2032146316730778004?s=46&t=1LAyoawP6LK1AbrwCvLGqQ

📰 Article • FixupX

We built RLM for coding. And it F*cking rocks. Swarm native agents are here to stay.

Today we are releasing slate. Slate is the *first* frontier agent in the wild to directly use a code environment for swarm orchestration. Slate can programmatically orchestrate and solve tasks,

Extract logos, colors, and images from any website URL. Open-source brand asset extraction toolkit.

https://openbrand.sh/

OpenBrand

https://www.cmux.dev/zh-TW

cmux

cmux — 專為多工處理打造的終端

面向 AI 程式 agent 的原生 macOS 終端。支援 Claude Code、Codex、OpenCode、Gemini CLI、Kiro、Aider 及任何 CLI 工具。

andrew gao (@itsandrewgao)

本质上是用 domain knowledge & glossary 精准命中 LLM 的内在知识储备，提醒我用 AI 做任何不熟悉的领域的事情前，都要先把概念和词汇掌握清楚，磨刀不误砍柴工

https://fixupx.com/itsandrewgao/status/2027579200635605058

推文中展示的网站是 https://component.gallery/

另一个领域词汇的网站 https://domainmaps.co/

🧵 Thread • FixupX

0:11

you can instantly 10x your vibecoded frontends by just learning what different ui components are called

ofc opus is creating generic slop, the only words you know are menu and button.

实践了一下，把今天用 claude code 做 plan review 的会话导出，然后写了个脚本将其生成为好看的 html, 再放到一个 repo 里自动构建网站发布到 github pages

https://reorx.github.io/agent-sessions/claude-code/usage-plan-review-and-durable-execution-idempotency.html

Claude Code's hidden conversation history (and how to actually use it) | @kentgigger

社交媒体上高赞的 Claude Code 使用教程/技巧文章我是一个都没点开过，但这篇我通过搜索发现的博客是真的值得一看

https://kentgigger.com/posts/claude-code-conversation-history

Uncommon Stack

Claude Code saves every conversation locally in ~/.claude/history.jsonl. Here's how to find old sessions, resume them, export them, and build a custom /history command.

hsliuping/TradingAgents-CN

🔗 https://github.com/hsliuping/TradingAgents-CN

#github

GitHub - hsliuping/TradingAgents-CN: 基于多智能体LLM的中文金融交易框架 - TradingAgents中文增强版

基于多智能体LLM的中文金融交易框架 - TradingAgents中文增强版. Contribute to hsliuping/TradingAgents-CN development by creating an account on GitHub.

shareAI-lab/learn-claude-code

🔗 https://github.com/shareAI-lab/learn-claude-code

#github

GitHub - shareAI-lab/learn-claude-code: Bash is all you need - A nano Claude Code–like agent, built from 0 to 1

Bash is all you need - A nano Claude Code–like agent, built from 0 to 1 - shareAI-lab/learn-claude-code

我觉得 vibe coding 时代，能够用来作为一个人开发能力参考的不再是代码了，而是与 coding agent 的对话记录，和最终产出的 spec 或 plan。前者比后者更重要，因为对话就是一个人思维过程的体现。

所以现在很需要一个工具能够保存和管理人与 code agent 的对话过程。这样我们日常开发时，遇到令人满意的 case，就可以把具有范例意义的对话标记存档——不仅可以给自己下次开发类似的功能时参考，也可以作为个人思维方式、AI 工具使用能力的体现。

所以未来我们的个人网站，不再是引导读者去看自己所写的代码，而是引导读者去查看一些 AI 对话。优秀的对话过程蕴含着一个人的认知、know-how，以及 AI 使用能力，这些都是新时代的开发者所应具备的。

为了避免我的 Mac mini 在长时间运行 OpenClaw 的过程中，出现我不清楚的状况，导致温度过高或者网络流量爆炸之类的危险状况，我写了一个脚本：

https://github.com/reorx/scripts/blob/master/mac-health

它的功能是将 Mac 当前的一些系统指标列出来，并根据配置文件中所设定的阈值进行检查。输出结果如下：

CPU Temp: 33.7 °C  ✅ 33.7 <= 55.0
Net RX: 0.04 Mbps  ✅ 0.0 <= 10.0
Net TX: 0.04 Mbps  ✅ 0.0 <= 10.0
CPU Load: 1.85 / 1.87 / 2.04  (19% of 10 cores)  ✅ 18.5 <= 50.0
Memory: 15.8 / 32.0 GB (57.3%)  ✅ 57.3 <= 80.0
Swap: 0.0 / 0.0 GB  ✅ 0.0 <= 0.0
Disk [/]: 395.5 / 460.4 GB (85.9%)  ✅ 85.9 <= 90.0

有了这个工具之后，我又让 OpenClaw 创建了一个 cron，在白天工作时间每小时运行一次这个命令，如果发现有任意指标出现异常，则推送消息给我。这样我终于可以安心睡个觉了。

scripts/mac-health at master · reorx/scripts

miscellaneous scripts I use. Contribute to reorx/scripts development by creating an account on GitHub.

jzOcb/context-doctor

🔗 https://github.com/jzOcb/context-doctor

#github

GitHub - jzOcb/context-doctor: 🧠 Visualize and diagnose OpenClaw context window usage — terminal, PNG, or JSON

🧠 Visualize and diagnose OpenClaw context window usage — terminal, PNG, or JSON - jzOcb/context-doctor

Stop Fixing Your Claude Skills. Autoresearch Does It For You

不知道 skills 的自更新使用哪個比較好，現在是把 auto research 的部分功能迭代到 hook 中結束後會自更新

也看到一個 YouTuber 用了類似的方式，但具體影片我也還沒看

https://youtu.be/qKU-e0x2EmE?si=ZiEfnyVeW9l_I3iB

https://github.com/karpathy/autoresearch

https://github.com/topoteretes/cognee

YouTube

🔥 Join Maker School & get customer #1 guaranteed: https://skool.com/makerschool/about
📚 Watch my NEW 2026 Claude Code course: https://www.youtube.com/watch?v=QoQBzR1NIqI
🎙️ The free SKILL.md: https://drive.google.com/drive/folders/14nUSxV8cpi5OI2OQxhBqyeuN92ERTMX1…

https://fixupx.com/hitw93/status/2032091246588518683?s=46&t=7IElwY_w1xvMGwebK0C6eg

FixupX

Tw93 (@HiTw93)