寫東西只為了抒發，一切隨緣不為流量

有事找 @finalfantasty

👻 個人博客 Blog: https://blog.pdzeng.com

👻 個人微博客 Blog:
https://daily.pdzeng.com/
熊貓隨口說
panda850819/murmur-voice 🔗 https://github.com/panda850819/murmur-voice #github
目前這個版本有幾個東西就還沒修正嘛例如說我目前都是壓著一個按鈕來講話未來可能也可以按一個按鈕然後讓它一段時間的自動去監聽我想要講什麼
然後再來就是後續的這些內容要不要透過 LLM 去排版這個我也還沒做
以及中英文夾雜的部分還有他為什麼會識別成簡體字的部分，我也還沒有想到比較好的解決方式
panda850819/murmur-voice

🔗 https://github.com/panda850819/murmur-voice

#github
GitHub
GitHub - panda850819/murmur-voice: Privacy-first voice-to-text for macOS and Windows. Local Whisper (Metal/CUDA) or Groq cloud…
GitHub - panda850819/murmur-voice: Privacy-first voice-to-text for macOS and Windows. Local Whisper (Metal/CUDA) or Groq cloud…
Privacy-first voice-to-text for macOS and Windows. Local Whisper (Metal/CUDA) or Groq cloud, with LLM post-processing. Built with Rust + Tauri 2. - panda850819/murmur-voice
github
我其實不知道要說什麼，太牛逼了
paulmckrcu/perfbook

🔗 https://github.com/paulmckrcu/perfbook

#github
GitHub
GitHub - paulmckrcu/perfbook: Is Parallel Programming Hard, And If So, What Can You Do About It?
GitHub - paulmckrcu/perfbook: Is Parallel Programming Hard, And If So, What Can You Do About It?
Is Parallel Programming Hard, And If So, What Can You Do About It? - paulmckrcu/perfbook
github
熊貓隨口說
之前有推薦過，但現在反悔了，應該是有資安風險的 1. 他基本上是開啟之後，就是全域監聽，包含文字可能也有剪貼簿 2. 如果你仍然想使用，可以將 Typeless 在你要講話的時候打開，剩下時間 cmd+q 關閉
現在很多人開源一些手搓版的，訂閱的錢當作虧了ㄅ，送他勒
不然就等他修正完一些安全性風險再看看了 = =
熊貓隨口說
https://www.typeless.com/
之前有推薦過，但現在反悔了，應該是有資安風險的

1. 他基本上是開啟之後，就是全域監聽，包含文字可能也有剪貼簿
2. 如果你仍然想使用，可以將 Typeless 在你要講話的時候打開，剩下時間 cmd+q 關閉
https://fixupx.com/fetalkpodcast/status/2021843045889581280?s=20
FixupX
前端輕鬆聊 (@fetalkpodcast)
Vercel 做了一個 agent md 和 skills 的測試，發現在 agent md 中放的 instructions 會比 skills 更高機率被你的 agent 觸發和遵守，他們發現因為 skills 在不特別講的前提下你的 agent 需要決定要不要用，而這又會根據你 skills 寫的條件如何而有所改變，如果 agent 覺得沒必要用就會跳過除非你特別指名要用哪個 skills。

他們這邊的範例是叫 agent 用最新的 next.js 文件，他們放了一些網址在 agent md 檔案中讓…
tinyfish-io/tinyfish-cookbook

🔗 https://github.com/tinyfish-io/tinyfish-cookbook

#github
GitHub
GitHub - tinyfish-io/tinyfish-cookbook: A collection of sample apps and recipes built with the TinyFish web agent. Open-source…
GitHub - tinyfish-io/tinyfish-cookbook: A collection of sample apps and recipes built with the TinyFish web agent. Open-source…
A collection of sample apps and recipes built with the TinyFish web agent. Open-source examples for you to learn & build! - tinyfish-io/tinyfish-cookbook
github
熊貓隨口說
Claude with Subframe https://fixupx.com/IrvinZhan/status/2022038492575154555?s=20
付費看起來也還是可以接受範圍內
Claude with Subframe
https://fixupx.com/IrvinZhan/status/2022038492575154555?s=20
Apple ZeroDay Update
https://nvd.nist.gov/vuln/detail/CVE-2026-20700

https://t.me/c/1987478432/37
PsiACE/bub

🔗 https://github.com/PsiACE/bub

#github
GitHub
GitHub - PsiACE/bub: Bub it. Build it.
GitHub - PsiACE/bub: Bub it. Build it.
Bub it. Build it. Contribute to PsiACE/bub development by creating an account on GitHub.
github
看完 真舒服

https://t.me/reorx_share/6521
Telegram
Reorx’s Forge
https://t.me/fming_weekly/655

一口气读完，享受到认知共鸣产生的多巴胺

做一个最小化的龙虾，你只需要下面几步：

1. 启动一个会写代码支持技能的 Agent（比如 Codex 或 Claude Code），让它写一个 startup 脚本，拉取 Telegram API 并用单次执行模式发给 Agent 自己。
2. 准备一个用这个脚本为启动脚本的 Dockerfile
3. 构建并运行这个 Docker 容器

把 claude code 作为核心引擎，实现一个 REPL…
https://herbertlui.net/recoverable-and-irrecoverable-decisions/
Herbert Lui
Recoverable and irrecoverable decisions - Herbert Lui
When you need to make an anxiety-inducing decision, ask yourself: What is the worst realistic outcome? Will you be able to recover from it? Let’s say you want to make a product and sell it, and you need to invest in inventory. The manufacturer needs you to…
https://dev.to/deeppressanalysis/open-source-radar-february-2026-agentic-ai-zero-trust-cicd-and-java-gc-2ikf
DEV Community
Open Source Radar (February 2026): Agentic AI, Zero Trust CI/CD, and Java GC
Open Source Radar (February 2026): Agentic AI, Zero Trust CI/CD, and Java GC
Open Source Radar (February 2026): Agentic AI, Zero Trust CI/CD, and Java GC Source: Analysis of Open...
Finding Alphas 導讀：量化交易的工廠操作手冊

https://hackmd.io/@UvzTVdP2Qxy6Q8UQ1U3A_g/H1uhkkjwZx
https://www.0xsid.com/blog/aidr
▶️ OpenClaw 太強了，磨合兩週後我的 AI 員工開始半夜自己加班了｜14 個排程完整公開ClawdBot） #youtube

https://www.youtube.com/watch?v=OB6o29I78rY
YouTube
OpenClaw 太強了，磨合兩週後我的 AI 員工開始半夜自己加班了｜14 個排程完整公開ClawdBot）
▸▸ 打造真正幫你省時間的 AI 系統，而不是更多工作 → https://www.guccidgi.com/join-ai-skool-yt

大部分人的 AI 助手，裝完第一天之後就再也沒打開過。

兩週前我裝了 OpenClaw，然後移除了。
但我沒有放棄——我把它部署到雲端主機，開始認真跟它磨合。

兩週後的現在，它從 5 個功能擴充到 14 個自動化排程。
每天晚上我去睡覺，它才開始最忙的時候——
半夜掃 GitHub 找 bug、深度閱讀我追蹤的 KOL 的文章和影片、
整理全平台數據、分析競品表現，…
youtube
https://dev.to/koushik_sen_d549bf321e6fb/repo-optimizer-i-let-a-kiss-ai-agent-optimize-itself-overnight-it-cut-its-own-cost-by-98-1ddi
DEV Community
Repo Optimizer: I Let a KISS AI Agent Optimize Itself Overnight. It Cut Its Own Cost by 98%.
Repo Optimizer: I Let a KISS AI Agent Optimize Itself Overnight. It Cut Its Own Cost by 98%.
No manual tuning. No architecture redesign. Just a plain-English instruction and a feedback...
https://dev.to/yannbuilds/i-built-a-cli-that-fails-your-build-on-broken-seo-425k
DEV Community
I built a CLI that fails your build on broken SEO
I built a CLI that fails your build on broken SEO
Last week, I spent two hours copy-pasting data from Screaming Frog into my CLI so Claude could...
Before
After
Back to Top