Bohan

外网就像一座围城，城外的人想进来，城里的人真会玩

[宝玉 @dotey]

Pi 两位作者的见解：

1. 代码即真相，代码不需要记忆系统，不需要 RAG，模型很擅长理解代码结构

2. Bash 工具足够用，Bash 类似于编程语言，可以任意组合；大部分时候没必要 MCP，skill + 脚本足够。

[雪踏乌云 @Pluvio9yte]

卧槽，夯爆了啊。
Notion、Figma、ClickUp这些大厂宣传片的动效直接被复刻成 Claude / Codex 能够调用的代码语言了：

106 张镜头配方卡（Shot Recipe Cards）：每张卡详细记录用途、能量级别、建议时长、参数、实现要点和常见坑。

161 条动态样片（覆盖 162 种样式）：可在线预览、搜索、筛选。

完整成片模板「Ink Press（墨压）」：36.2 秒、1920×1080、30fps、10 个镜头的纸墨琥珀风格产品宣传片，已包含 2.5D 页面相机、转场、字幕和配好的 SFX。

可复用组件：2.5D 页面相机、闪切、数字滚动、字幕、音效等。

完整制作方法论：从素材采集、风格定调、分镜、声音设��、节奏卡点到最终验收的全流程文档。

开源地址
https://t.co/DpYv77UWgg

[Bohan @Bohaaan]

@CryptoJHK 直接泄露用户数据，不可思��

[Bohan @Bohaaan]

@jiayuan_jy multica 非常好用👍我写了一个 skills 方便其他 agent 操作 issue、skills、附件等等。

https://t.co/Yuzr2rUR5x

#multica #skills

[Bohan @Bohaaan]

@bulletonbible 创意不错👍试了一下好像卡死了，一直没有 json 输出

[Bohan @Bohaaan]

Taking longer than expected ... @cursor_ai this is a bug!

[Bohan @Bohaaan]

@jiayuan_jy skills 貌似很难安装，好多都没有这个 skill.md

[Anthropic @AnthropicAI]

New on the Anthropic Engineering Blog:

How we use a multi-agent harness to push Claude further in frontend design and long-running autonomous software engineering.

Read more: https://t.co/HWvmXk1ykn

[Bohan @Bohaaan]

视频压缩工具 - 在线压缩视频减小90%文件大小 | VideosKit - VideosKit https://t.co/czgn2eWgiE #VideosKit #tools #productivity #developer

[Bohan @Bohaaan]

nice👍

[VideosKit @videoskit]

🚀 Meet VideosKit — free browser-based video tools that respect your privacy

✅ 6 tools: compress, trim, crop, convert, GIF, info viewer
✅ 100% in-browser
✅ No uploads, no signups, no tracking

Try it: https://t.co/pG6Y1BErgJ

#opensource #webdev #video

[Bohan @Bohaaan]

@jaredliu_bravo @openclaw 最近又准起来了，之前让他每小时心跳发一次消息，一天也就发 3-4 次🙄

[Bohan @Bohaaan]

这个好

[Vercel Developers @vercel_dev]

You can now write chat bot logic once, and deploy across Slack, GitHub, Teams, and Discord.

We're open-sourcing a new Chat SDK, a unified TypeScript library to build for many chat platforms with a single codebase.

Now available in public beta.
https://t.co/UICfmvl9BR

[Bohan @Bohaaan]

🦞openclaw × remotion！自动写小红书笔记并剪辑视频，这个方案有的玩吗？

[Bohan @Bohaaan]

@HiFrey 浙江再加上秋假，半个月了

[Okara @askOkara]

@AnthropicAI

[Bohan @Bohaaan]

人类早期驯服🦞龙虾的珍贵影像资料，个人拿 openclaw 做内容还是挺有潜力

[Bohan @Bohaaan]

@chengfeng240928 讲的很清楚👍

[Bohan @Bohaaan]

@ahshengchen glm5 还挺慢

[Alex Xu @alexxubyte]

Cursor recently shipped Composer, its agentic coding model, and shared that the agent can be ~4× faster! We worked with the Cursor team, particularly @leerob, to understand how the system is put together, and what drives the speed.

A coding agent is a system that can take a task, explore a repo, edit multiple files, and iterate until the build and tests pass.

Inside Cursor, a router first picks a suitable coding model (including Composer) to handle the request.

The system then starts a loop: retrieve the most relevant code (context retrieval), use tools to open and edit files, and run commands in a sandbox. Once the tests pass, the task is complete.

Cursor uses three key techniques to keep this loop fast:

1. Mixture-of-Expert (MoE): A sparse MoE architecture activates only a subset of model weights per token.

2. Speculative decoding: a smaller model drafts multiple tokens at once, then a larger model verifies them in parallel to reduce latency.

3. Context compaction: summarize older steps and keep only the active working set so the prompt stays relevant and short as iterations continue.

Full newsletter here: https://t.co/WxnIBuL3Ow