大家好，我叫Casey，这其实是我最喜欢的Linux发行版，主要是因为Deepin在美学和用户友好设计方面超越了其他所有发行版，而且在不限制开发者或者那些需要编程或进行更复杂任务的用户的情况下做到了这一点。

我想与其他Deepin OS用户分享一些有用的资源（工具、指南），帮助打造一个更加AI中心化的系统，或者说其他你想称之为的东西，希望这不会浪费时间，哈哈。

好吧，以下是可以在用户友好的界面中本地运行的图形化AI LLM模型：

• msty.ai 应用 - https://next-assets.msty.studio/app/beta/linux/MstyStudio_amd64.deb（直接的.deb链接）或 https://next-assets.msty.studio/app/beta/linux/MstyStudio_x86_64.AppImage（AppImage链接）
  （备注：为了让使用AppImage应用更加方便，强烈推荐你从GitHub获取AppImage Launcher .deb，或者通过Flatpak安装GearLever，使用起来非常方便）
• Cherry Studio是另一个用户友好的图形化LLM应用：https://www.cherry-ai.com/
• 我最喜欢的应用，它可以让你轻松点击安装几乎所有最好的生成型AI项目，例如音频克隆、照片和视频编辑、浏览器自动化、openwebui等，安装完成后，只需点击一次就能启动你想要的项目，直接的.deb链接是：https://github.com/pinokiocomputer/pinokio/releases/download/3.9.0/Pinokio_3.9.0_amd64.deb

哦，关于编程（Agentic），比如你希望从主要的LLM提供商（如OpenAI、Google、Claude AI/Anthropic）中获取你的编程代理，首先需要安装nodejs、npm、nvm和npx，执行以下命令：

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh
$HOME/.nvm/nvm.sh
nvm install 22
nove -v   # 这一步验证安装的版本号是否正确

然后：

• Claude Code：npm install -g @anthropic-ai/claude-code
• Gemini CLI：npm install -g @google/gemini-cli
• OpenAI Codex：npm install -g @openai/codex

另外我发现一个非常有用的工具，来自AWS，注册一个免费账号并安装Amazon Q Developer CLI，有AppImage应用以及一个实际基于Ubuntu的压缩归档版本，我认为后者更方便，只需解压，使用命令 chmod +x ./install.sh，然后 bash ./install.sh进行安装。安装完成后，刷新终端/命令行，输入 source ~/.bashrc并按回车，输入字母q后就可以直接用自然语言输入命令，比如：列出当前目录下的所有文件、修改系统配置、增强系统性能设置等等。

最后，如果你想尝试一些具有丰富免费配额的Agentic编码模型，并能链接到你的GitHub仓库以自然语言提交要做的事情，可以试试Google的Jules，访问jules.google.com即可使用，只需要一个Google账号。它可以直接在GitHub上对你现有的项目进行编码和优化，不过我个人找到的最好的工具其实是factory.ai、openhands和MiniMax。

哦，还有，我忘记提到的是很酷的Agentic AI浏览器：（在浏览器内执行多步自主网站交互任务的代理）

• 有：browserOS、fellou、以及Perplexity发布的Comet浏览器，另一个即将发布或某些地方已经有的就是Opera的Neon浏览器。

关于mcp（模型上下文协议和工具，使其易于提供给代理执行各种任务），我推荐使用Docker的图形化桌面应用，不过也有很多可以通过npm安装并快速运行的mcp工具。

希望这些对大家有所帮助。随时评论回复，如果你有任何自己的建议或推荐的工具，别忘了分享给我和其他人。谢谢大家！

Hi everyone my name is Casey, this is actually my favourite Linux distro, primarily because Deepin have outdone every other in terms of aesthetics and a user friendly design, but done so without limiting developers or those wanting to do coding or more complex tasks.

I wanted to share several useful items with other Deepin OS users (resources, guides) to make it a more AI centric setup? Or whatever you wish to call it, hopefully not a waste of time though lol

Ok for graphical AI LLM models you can run locally in a user friendly interface:

msty.ai app - https://next-assets.msty.studio/app/beta/linux/MstyStudio_amd64.deb (direct .deb link) or https://next-assets.msty.studio/app/beta/linux/MstyStudio_x86_64.AppImage for the (AppImage) (Note: To make using AppImage applications easier to use I highly recommend you get AppImage Launcher .deb from GitHub, or via Flatpak get GearLever, makes it extremely easy to install locally)

Cherry Studio is another user friendly graphical llm app: https://www.cherry-ai.com/

My favorite app which inside of it allows click and install user friendly setup of almost all the best generative AI, eg. audio cloning, photo and video editing, browser automation, openwebui , and many more projects you can launch just by clicking once it finishes installing the ones you want, the direct .deb link is https://github.com/pinokiocomputer/pinokio/releases/download/3.9.0/Pinokio_3.9.0_amd64.deb

Oh and for coding (Agentic) (eg. where you have your coding agent from for example the main (main in the west) llm providers such as openai, google, claude ai/anthropic.) , firstly get nodejs , npm, nvm, npx by runnning the following:

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh

$HOME/.nvm/nvm.sh

nvm install 22

nove -v # This verifies the version number ensuring it installed properly

Then:

for Claude Code: npm install -g @anthropic-ai/claude-code

for Gemini CLI: npm install -g @google/gemini-cli

for OpenAI Codex: npm install -g @openai/codex

Another one I found to be very useful is from AWS, sign up for a free account and install Amazon Q Developer CLI, there is an AppImage app and an actual ubuntu based distro compressed archive which is easier or better I believe, just unzip it, chmod +x ./install.sh , then bash ./install.sh to install it and to launch it simply refresh the terminal/shell by typing source ~/.bashrc hit enter, type the letter q and simply type in natural language, eg. list all files in current directory, modify my systems configurations, enhance system performance configurations, or whatever you wish.

Lastly, if you want to try some agentic coding models with generous free quotas and ability to link your GitHub repositories select which one to work with issue in natural language what you want done try Google's Jules, jules.google.com , just need a Google account to use it, it can perform coding and enhancements on your existing projects directly onto GitHub, although the best one I have found is actually factory.ai , openhands, and MiniMax.

Forgot to also add this which is awesome, agentic AI browsers: (multi step autonomous website interacting agents within browser)

There is: browserOS , fellou, and Perplexity released their Comet browser, another one coming or available in some places is Opera's Neon browser.

For mcp (model context protocol and tools make them easily available to agents to use and perform an array of tasks) I recommend Docker's graphical desktop app, but there are many nodejs (npm install....) mcp items easy to install and get running.

Hopefully that helps people out. Feel free to comment reply , add any of your own personal suggestions or recommendations for me and others to try that aren't listed. Thank you.