If you’ve switched to Opus 5.5 but still have a bunch of unused Codex resets, here’s how I’ve been using mine.
I let Opus orchestrate Codex through Herdr, T3, Hermes, OpenClaw, whatever works. I also added instructions to my personal CLAUDE.md telling Opus to hand off backend work and bug fixes to Codex, while keeping the creative work and anything involving design or aesthetics for itself.
The trick is making sure Opus gives Codex really specific instructions, because I’ve found that vague requests don’t work nearly as well. I also require Opus to visually verify the results and, if something is wrong, tell Codex exactly what to fix rather than taking over, although sometimes Opus does need to step in and handle things itself when Codex just isn’t getting it.
I still get the creative judgment I like from Opus, my Claude usage goes further, and those Codex resets aren’t going to waste.
Holy sh*t, someone just open-sourced the complete motion design stack for Claude Opus 5.5, and I am going to tell you about the best repositories that nobody is talking about
How does this work:
3b1b/manim - 65,400+ stars
The mathematical animation engine behind 3Blue1Brown. It enables Opus 5.5 to write programmatic, frame-perfect 2D scenes and dynamic diagrams directly in Python without visual template drift
https://t.co/rIg01m4Kjz
microsoft/playwright - 66,200+ stars
Headless browser automation framework that forms the capture backbone. Enables Opus 5.5 to orchestrate Chromium frame-by-frame at deterministic 60 FPS under a seek(t) harness
https://t.co/L9LJp33Rbd
juliangarnier/anime - 48,500+ stars
A lightweight JavaScript animation engine with complex sequencing and SVG path morphing. Opus 5.5 uses it to write mathematical easing curves and layered micro-interactions directly into canvas
https://t.co/F2re96YRuQ
pixijs/pixijs - 43,200+ stars
The fastest 2D WebGL rendering engine for the web. Provides hardware-accelerated canvas pipelines so Opus 5.5 can generate particle systems, complex shader transitions, and UI motion at native refresh rates
https://t.co/bK2VOynazV
m-bain/whisperX - 16,800+ stars
Word-level forced audio alignment powered by Whisper. Turns speech audio into millisecond-accurate timestamps so Opus 5.5 can sync kinetic subtitles and dynamic captions to every syllable
https://t.co/SdcXQYOPdo
remotion-dev/remotion - 22,400+ stars
The core programmatic video engine written in React. Enables Opus 5.5 to treat entire launch films and UI walkthroughs as composable React components rendered straight to MP4
https://t.co/895JHjBqFY
theater-js/theater.js - 12,100+ stars
A programmatic motion design and visual timeline orchestration suite for Three.js and HTML. Gives Opus 5.5 a multi-track timeline substrate to animate cameras and lighting in 3D space
https://t.co/GnGbClWhOu
greensock/GSAP - 18,900+ stars
Industry-standard high-performance animation engine. Handles complex timeline chaining and SVG morphs, giving Opus 5.5 sub-millisecond precision over DOM layouts and camera pans
https://t.co/HizIemBKwP
airbnb/lottie-web - 29,800+ stars
Mobile and web vector animation runtime. Renders pure code and After Effects vectors natively without relying on heavy compressed video files
https://t.co/XXgZSxmswC
video-db/videodb-python - 2,800+ stars
Programmable video infrastructure that enables Opus 5.5 to ingest raw footage, perform semantic frame searches, and automate B-roll insertion directly inside its agent workflow
https://t.co/PJKKInZzl2
You will definitely find what you need
I recommend adding this to your bookmarks ↓
Pokémon TCG illustrator Haru Akasaka (@Haru_Akasaka) stopped by MAG booth S41 at LahLahLand yesterday. Thank you for visiting, Haru. Haru is a guest of the CardsLAH Card Party this weekend.
Pack a submission in this order: penny sleeve, then Card Saver. Not a top loader. Card Saver = semi-rigid holder. Don't clean the card; leave the surface as it is. Packing guide on our blog: link in bio.
Needed a dynamic AI orb for some tests, so I ported this one to React Native
Try it yourself in Native Springs Shaders: https://t.co/MxOdSkvOVk
Want more orbs like this, ready to drop in?
WERE BRINGING CODEX COMPUTER USE TO HERMES.
If you haven’t used codex computer use yet, it’s 2-4x faster and cheaper than regular browser control.
And we’re reverse engineering it for Hermes…
this is Alan’s way!
¡UNA STORE DE JUEGOS DIRECTAMENTE EN PS5!
Orbit Store permite navegar y descargar contenido desde la propia consola.
◽ 581 juegos en el catálogo inicial
◽ Descargas directas
◽ Pausar y reanudar
◽ USB/almacenamiento
◽ Control desde el móvil
⚠️ Beta
https://t.co/YGkpEgVjzp
1 tahun lepas, this might be hard for simpleton like me to setup this.
Tapi sekarang, you just subscribe pakej rahmah OVH atau netcup < RM30/bulan, masukkan endpoint then suruh je your agent buatkan. BYOK sedia ada.
Margin bisnes ada, tapi paper thin. Theres no moat anymore.
browserbase is dead
you can spawn 2k concurrent 24/7 VMs on https://t.co/7XB0X7XNXc and spin up to 29M browser sessions 3 min long each, per month, for just 26k
vs 155k on them
used it to scrap 366 products & their funding for https://t.co/GmrJuSKcJX
One Piece 3on3 Regionals at Hextar World, Empire City, PJ today. Good luck to everyone playing. Protect promos and prize cards: penny sleeve, then a Card Saver (semi-rigid holder).
We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:
Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better:
Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:
Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:
Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!
In summary:
- As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.
- Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.
New component for @expo — Leaderboard.
Entries ranked by a value, with the top three on a podium.
Suggested by @mikeccrane. Thank you. 🙏
Part of PanelUI open-source, copy-paste, you own the code.
https://t.co/fwcNUq4h8T
#ReactNative#OpenSource#expo
Introducing React Native Deploy: free unlimited mobile cloud bluilds managed by you through Github runners.
~/ npm install -g @react-native-feel/deploy
- Share preview builds with a QR code
- Send to TestFlight
- eas.json compatible
https://t.co/vSNT6o0dmJ
I'm BEGGING you to look up what YouTube pays per 1,000 views in different niches before you pick what to make videos about
This is the single most important number in the entire YouTube business model and almost nobody checks it before starting
Entertainment: $2-4
Gaming: $3-5
Vlogs: $4-6
Reactions: $3-5
Now look at the niches nobody wants:
Medicare supplements: $44
Estate planning: $38
Commercial insurance: $35
Business tax strategy: $32
HVAC repair: $28
Assisted living: $26
Septic and well water: $24
Farm equipment: $22
That's not a typo. $44 per thousand views vs $3 per thousand views. The exact same platform. The exact same algorithm. The exact same upload button
A gaming channel needs 1,000,000 views to make $3,000
A Medicare channel needs 68,000 views to make $3,000
And the Medicare channel has maybe 12 competitors in the entire United States while the gaming channel has 45,000
But this is where it gets stupid:
The $44 CPM is just the ad money. The viewers in these niches are about to spend $5,000-$200,000 on the thing you're explaining to them. The ad revenue is a tip. The real money is what happens after they watch
A roofer with 3,000 subscribers who answers "how much does a new roof cost" gets calls from homeowners about to spend $14,000. Every video is a free sales rep that works at 3am
A financial advisor with 5,000 subscribers who answers "should I do a Roth conversion" gets appointments from retirees managing $400,000+ portfolios. One client is worth $5,000-$15,000 a year in advisory fees
An HVAC company with 2,000 subscribers who answers "heat pump vs furnace" gets calls from homeowners about to spend $8,000-$15,000. Those viewers aren't subscribing to an HVAC channel. They're panicking because their furnace died in January and they need someone who knows what they're talking about
The leads close at 2-3x the rate of ad leads because the buyer watched you explain their exact problem for 15 minutes before they called. The trust was built before the conversation started
How to find these niches yourself in 10 minutes:
1 → Open Google Keyword Planner (free with a Google Ads account you never have to fund). Type any service. Sort by "Top of page bid - high range." That column shows what advertisers pay per click. $50-200 per click = that niche pays $20-44 CPM on YouTube. The correlation is almost perfect
2 → Search that topic on YouTube. Count how many serious channels answer it. If there are fewer than 15, you have almost no competition for an audience advertisers are paying $50-200 a click to reach on Google
3 → Check the comments on the few channels that exist. Every question asked is a video idea nobody made yet. Every complaint is a positioning opportunity
The uglier the niche sounds the emptier it is and the higher it pays. This has been true for 8 years straight and nobody talks about it because making videos about septic systems and Medicare won't get you likes in a group chat
I build channels in these niches. The ones that sound like pamphlets in a doctor's waiting room. Because that's where the checks clear
(link in bio)
Finally found my ideal remote agent setup.
Hetzner + Ocra
Orca runs on my laptop, iOS and Hetzner server.
I can start a project at home and continue in the go, without being afraid it breaks anything.
All are connected via Tailscale so the server is secure + can serve me preview builds no matter where I am.
I’m building a wrapper around it now that gives me a kanban task board and many more learnings from my previous AgentOS.
Fun!
Excited to release the @github async merge API today!
If you (or your agents) are merging PRs programmatically, this is for you 🧵
https://t.co/F7r7AAFV9k
use e2e alongside @poteto's verification skill and you have a reliable workflow for mobile development.
the only drawback of developing a feature using pstack was the verification was pretty much unit testing,
I had to setup and create a whole maestro flows to test the mobile apps, but it was not practical, it took long time
i have tried @agent_device@swmansion's agrent (alongside with @appandflow's stim), and it was great but I felt like (agent-device and argent are basically for development maybe I'm mistaken) but just tested out e2e alongside pstack and that's what verification skill expected.
still trying to figure out the most effiecent way around the agentic workflow, but it's already looking promising
- pstack for nearly everything
- stim for orchestrating and warmup sims
- @agent_device mostly for querying the simulator
- argent for querying and mutating the sim (found it to be a little bit faster from @agent_device
- e2e to make sure that the feature just works
don't also sleep on https://t.co/CkIbrkhaI9 also by the one and only @o_kwasniewski
SimSlim has to be one of the best discoveries for mobile developers working with the iOS Simulator.
For the first time, my Simulator is running without making my PC fan go crazy, heat up, or slow down. 😭🔥
https://t.co/8uUoh43WkM