Pinned Tweet
As an AI hardware founder in Shenzhen, I sift through tons of smart devices daily. Most of them are pure noise.
So I spent the last 2 months putting together almost every single AI hardware project on the market that is actually worth seeing.
Click into any product to get its ub repo.
Enjoy! 👇
So I spent the last 2 months putting together almost every single AI hardware project on the market that is actually worth seeing.
Click into any product to get its ub repo.
Enjoy! 👇
88
149
1,531
81,373
Everyone is losing their minds trying to get the new iPhone Duo. I couldn't get one, said screw it, and made my own out of an old Nintendo DS. It's open-source btw
IPHONE DUO BUT MADE IN SHENZHEN🍎🎮
IPHONE DUO BUT MADE IN SHENZHEN🍎🎮
L’émulation Nintendo DS n’était pas dans mon bingo des choses qui rendraient l’iPhone Duo trop cool, mais je m’avoue conquis 😱
34
654
5,232
313,307
Built a 3D web simulator for anyone who doesn't have the console. Try it out in your browser👉josephzhou.me/3ds/
0
1
13
2,374
Don't have a DS lying around? I also made a 3D web version so you can play with it right in your browser👇
Everyone is losing their minds trying to get the new iPhone Duo. I couldn't get one, said screw it, and made my own out of an old Nintendo DS. It's open-source btw
IPHONE DUO BUT MADE IN SHENZHEN🍎🎮
3
8
57
2,747
JosephZhou retweeted
Claude Opus 5.5 - Manga Edit Prompt Share
Saw a Makima edit while scrolling and thought I'd give it a try with Opus using one of my Makima videos.
It created this entirely with code, using just the video from the quoted post and a music file as references. Definitely gonna try his!
Prompt in the replies. 👇
Saw a Makima edit while scrolling and thought I'd give it a try with Opus using one of my Makima videos.
It created this entirely with code, using just the video from the quoted post and a music file as references. Definitely gonna try his!
Prompt in the replies. 👇
MiniMax H3
Camera control with Makima!
You can use your own character. Check the prompt below!
26
30
412
19,327
なんか日本でバズってる…！見てくれてありがとうございます🙏
いつもこういう変なUIやガジェットを勝手に作って遊んでます。面白いと思ったらぜひフォローしてください🙌
x.com/openmys/status…
いつもこういう変なUIやガジェットを勝手に作って遊んでます。面白いと思ったらぜひフォローしてください🙌
x.com/openmys/status…
Groki Bot: a desktop body for Grok Bot.
Hardware: M5Stack CoreS3 (ESP32-S3)
- Touch it, and it reacts with emotions.
- Control Grok Bot by voice. Anything you'd ask it in
the app, you can just say out loud.
- It turns to follow your face as you move.
Watch it work: I ask out ub issue.
Grok Bot goes to GitHub, finds it, and the robot reads me
what it's about. I say "mark it done", and Grok Bot closes
the issue. No keyboard, no screen.
Open source soon.
@bot @romanugarte_ @spacexai @elonmusk
0
2
30
5,431
MAKE 3DS GREAT AGAIN
GTA V on a A̶Y̶N̶ ̶T̶h̶o̶r̶ 3DS XL
Running at 60fps with frame gen, full second screen support, touch weapon wheel and map on bottom screen.
All from a reverse engineered Xbox 360 ISO, meaning the games only 17GB.
Been working on this since before the Web Assembly version
1
0
21
1,373
Cooking up something big🏊 Currently working on turning every piece of opensource hardware on this site into an interactive prototype.
As an AI hardware founder in Shenzhen, I sift through tons of smart devices daily. Most of them are pure noise.
So I spent the last 2 months putting together almost every single AI hardware project on the market that is actually worth seeing.
Click into any product to get its ub repo.
Enjoy! 👇
2
0
29
1,605
虽然这个时代我可以用定时任务让agent定时总结我的所有社交平台的活动，以及每半小时跟我汇报一下互动情况。但是我就是控制不了自己去手动刷新，对于ADHD创作者来说，社交平台通知图标右上角的气泡简直就是刺激多巴胺最直接的来源。
4
0
4
759
We plugged Google Research’s Harness² into JEV 1.14.
Zero weights touched, yet driving massive gains on vision benchmarks round after round.
Under the hood, the system:
⚡️ Rewrites its own prompts
🛠️ Builds its own tools
🧠 Distills its own experience
JEV-Harness² repo
Zero weights touched, yet driving massive gains on vision benchmarks round after round.
Under the hood, the system:
⚡️ Rewrites its own prompts
🛠️ Builds its own tools
🧠 Distills its own experience
JEV-Harness² repo
Google just open-sourced Harness².
(turn any harness into a self-improving system)
a harness is everything wrapped around the model, like the system prompt, tools, memory, skills, and subagents. today a person designs it once, and the agent runs every new task with that same lets the agent refine that setup for each new task on its own, with the model weights left untouched.
on every task, the agent proposes edits to its harness, runs a few candidate versions, and composes a better harness by comparing how those runs behaved.
because it learns from the differences between runs, it needs no correct answer or grader for the task.
that covers a single task. Harness² also carries what it learns from one task into the next.
after each task, it saves the harness it settled on along with the lessons that held up across the runs. the next time it refines a harness, it starts from those notes instead of from scratch.
this means every finished task makes the next round of edits more informed.
Google tested this with Gemini 3.5 Flash and GLM-5 on legal, office, and professional work benchmarks. one round of edits was enough to beat their original harnesses, and giving the agent more rounds pushed scores higher.
those extra rounds can go wide or deep. the agent either tries several candidate harnesses at once or keeps refining a single one step by step, using OpenCode or Codex to carry out the work.
you can find the Harness² repo here: github.com/google-researc…
I wrote about a similar idea for agents already running in production, where the harness should repair itself after a failure instead of waiting for someone to debug the trace by hand.
that piece walks through how @Cometml's Opik (open-source) closes this loop, going from a failing trace to a proposed fix to a regression test that stops the same failure from coming back. it's quoted below.
64
288
2,754
176,174
The architecture diagram is now open-source👇
x.com/openmys/status…
x.com/openmys/status…
Got a lot of questions after this post on how the demo video was actually made.
Today, I'm open-sourcing cinetty — the architecture diagram Skill that has gotten me over 1M views on X.
Think of it as an After Effects template library built specifically for the terminal. It , knowledge structure, or complex workflow into a dynamic TUI interface in one click.
No need to manually tweak coordinates or write animation code. It comes with 6 built-in templates and native support for Claude Code and Codex.
Specifically optimized for Chinese (CJK) character rendering and alignment in terminal diagrams.
Just feed it as a Skill to your coding agent: let the agent read your repo, populate the content, validate it, and run a cinematic architecture animation right in your terminal.
Now it's your turn to create your own TUI digital art.
0
0
0
132
Got a lot of questions after this post on how the demo video was actually made.
Today, I'm open-sourcing cinetty — the architecture diagram Skill that has gotten me over 1M views on X.
Think of it as an After Effects template library built specifically for the terminal. It , knowledge structure, or complex workflow into a dynamic TUI interface in one click.
No need to manually tweak coordinates or write animation code. It comes with 6 built-in templates and native support for Claude Code and Codex.
Specifically optimized for Chinese (CJK) character rendering and alignment in terminal diagrams.
Just feed it as a Skill to your coding agent: let the agent read your repo, populate the content, validate it, and run a cinematic architecture animation right in your terminal.
Now it's your turn to create your own TUI digital art.
Today, I'm open-sourcing cinetty — the architecture diagram Skill that has gotten me over 1M views on X.
Think of it as an After Effects template library built specifically for the terminal. It , knowledge structure, or complex workflow into a dynamic TUI interface in one click.
No need to manually tweak coordinates or write animation code. It comes with 6 built-in templates and native support for Claude Code and Codex.
Specifically optimized for Chinese (CJK) character rendering and alignment in terminal diagrams.
Just feed it as a Skill to your coding agent: let the agent read your repo, populate the content, validate it, and run a cinematic architecture animation right in your terminal.
Now it's your turn to create your own TUI digital art.
We plugged Google Research’s Harness² into JEV 1.14.
Zero weights touched, yet driving massive gains on vision benchmarks round after round.
Under the hood, the system:
⚡️ Rewrites its own prompts
🛠️ Builds its own tools
🧠 Distills its own experience
JEV-Harness² repo
5
6
38
2,842
JosephZhou retweeted
Finally full DOOM on a Vape😅
Source:
github.com/atc1441/Vape_D…
Since the previous custom firmware was only Screen Sharing it this was finally needed!
Only possible by Arturs impressive work on making DOOM go bzzz on 32kb RAM
Source:
github.com/atc1441/Vape_D…
Since the previous custom firmware was only Screen Sharing it this was finally needed!
Only possible by Arturs impressive work on making DOOM go bzzz on 32kb RAM
102
1,111
9,746
544,521
So happy our canvas UI helps you get into the flow exploring open-source hardware!
Want results without the flashy animations? Witness Shenzhen Speed ⚡️
All devices, prices & repos in 1 synced link for your Agent—optimized from 89k to 5,800 tokens!
You're welcome, have a great day! 😎
Want results without the flashy animations? Witness Shenzhen Speed ⚡️
All devices, prices & repos in 1 synced link for your Agent—optimized from 89k to 5,800 tokens!
You're welcome, have a great day! 😎
As an AI hardware founder in Shenzhen, I sift through tons of smart devices daily. Most of them are pure noise.
So I spent the last 2 months putting together almost every single AI hardware project on the market that is actually worth seeing.
Click into any product to get its ub repo.
Enjoy! 👇
2
0
16
2,228
Huge thanks for all the support! It makes me so happy to see our canvas-style UI helping you get into the flow state while exploring open-source hardware. 🙏✨
However, in this day and age, everyone's time is precious. For my friends thinking: "Skip the flashy animations, just give me the results!" — I decided to let the world experience some ShenZhen Speed! ⚡️
I’m opening up the entire dataset directly for your AI Agents.
• All-in-One: Every device, price, and open-source repo in ONE link, always kept in sync.
• Token-Friendly: Don't worry about token limits. I heavily optimized the data from 89,000 tokens down to just 5,800!
Just feed this straight to your Agent:josephzhou.me/ai/agent
You're welcome, and have a wonderful day! ☕️😎
However, in this day and age, everyone's time is precious. For my friends thinking: "Skip the flashy animations, just give me the results!" — I decided to let the world experience some ShenZhen Speed! ⚡️
I’m opening up the entire dataset directly for your AI Agents.
• All-in-One: Every device, price, and open-source repo in ONE link, always kept in sync.
• Token-Friendly: Don't worry about token limits. I heavily optimized the data from 89,000 tokens down to just 5,800!
Just feed this straight to your Agent:josephzhou.me/ai/agent
You're welcome, and have a wonderful day! ☕️😎
1
1
7
1,270