Yes. Considering how much they dropped the ball lately, before Opus 5.5, I’m just waiting for the fumble.
When it comes to the business side of things, Anthropic has butter fingers.
Just hit 2,800 followers.
Seriously appreciate everyone who’s followed along, liked, replied, shared something, or just quietly read my posts.
I’m always trying to make the content better and more useful, so thank you for sticking around.
Onward.
An agent skill can look harmless. Sometimes all you see is Markdown.
That file can still hide instructions that push an agent to leak data or run commands you never intended.
NVIDIA's SkillSpector scans for prompt injection, data exfiltration, and other risks before installation.
Scan first. Trust after.
I’m kind of blown away by how far Codex can take App Store screenshots with a simple compositing and art-direction pass.
You give it a normal App Store shot, a clear visual direction, and it comes back looking absolutely cracked.
One of the great beauties of AI that few people talk about is no longer having to debug code for hours on end.
AI can do that now, and it does it well.
2,500 of you are following this account now, which is pretty incredible!
I started posting mostly to document what I was learning and building.
Somewhere along the way, it turned into a community of people I learn from, too.
Thank you for sticking around.
A couple days ago I posted that I had no clue how to use tryanim8. I mostly work backend, and Figma is not one of my strengths.
So I set aside a couple hours and taught Hermes Agent how to use it.
The backstory made the project even better.
On August 6, @adrianabelarde was experimenting with Seedance 2.0 and vectorization in Paper for the Clucky launch video. Codex helped clean up artifacts. He still needed a better way to morph between frames.
Two days later, he had his own vectorizer and smarter frame animation working. On August 19, he released tryanim8 as the tool behind the Clucky launch video.
It could turn images into editable SVGs and animate them in one workspace. The editor exported an MP4 video and editable SVG files. It also supported Lottie. Three days after launch, he posted that anim8 had reached $507 in monthly recurring revenue.
This is the kind of builder story I love. A real project created a real problem. Solving it produced another product.
I wanted Hermes to learn the editor deeply enough that future animation projects would have a repeatable starting point.
We mapped the editor and learned the .anim8 project structure. Then we documented the browser and export quirks before building an original 12-second animation called Pogo Pop Lab V2.
The final project had:
78 layers
8 groups
60 animated layers
1,101 keyframes
Hermes exported the video and checked exact transition frames. It found group-origin and layer-order problems. After fixing them, Hermes verified that tryanim8 preserved the complete editable project.
That work now lives in a reusable Hermes skill.
My next app launch, product demo, or animation experiment can start with a tested process instead of another blank canvas.
A couple hours spent teaching an agent that records its progress can keep paying you back across future projects.
Another 83 followers today! That is pretty rad!
Analytics says 83 follows, 4 unfollows.
I'll take that as a win!
What are you doing to grow on X on a daily basis?
Me? Just posting and replying more consistently.
Rest of the world on Glassdoor:
“AI isn’t life-changing, I just check other people’s AI work now”
Us:
Already treating that entire skillset as temporary legacy infrastructure.
This is peak denial before the flood.
We are so early it’s just crazy when you think about it.
I created an LLC for products and services I'm going to offer based on my experience, and even some inexperience that AI tools can help me with along the way.
I built a neon pink landing page for Node Lab Studio just to establish something to celebrate this little win.
It turns the Harbor & Cross website glow style into a black and pink look using two radial glows and a masked image glow.
Don't get me wrong, I love my job and the work that I do, but I don't want to one day look back and regret not having tried to start my own thing, either.
2,000 of you are following along now.
No big content strategy. Mostly just me at a terminal after work, building things and sharing what I learn along the way.
Every reply has made me better at this.
Really appreciate you all.