🍯 Edgar (@EdgarsHQ) header
🍯 Edgar (@EdgarsHQ) profile picture

Always cooking

Europe
Joined July 2019
Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)Photo by 🍯 Edgar (@EdgarsHQ)
🍯 Edgar retweeted
Meet the new Claude mobile app! So much craft and polish across the app, including a new tab bar. Huge shoutout to @daniel__designs, @khalilcader, @pranathiperii, Andrew, and Graham for all their work across iOS and Android. Rolling out to all users, starting today!
208
189
41
4,928
313,608
If you think @Apple can pull off AI - try search in their Mail App
7
Fable 5.2 was too fun… Shipped several dozens PRs over the weekend for agentx
14
Lately I feel rusty with pure UI work because of prebuilt components and little challenges, but this project makes it interesting to give a try and play myself, dive deeper
SPEC-01/HARDWARE Themed components built with @Vercel UI SDK spec-01.vercel.app Sound on
20
agentx is getting a look:
3
At the speed of light
What's new in Grok Bot 00:00 🌁 00:10 Phone calls 01:02 Share sheet 01:27 X cards 01:40 Shopping deals 01:58 Gifts 02:26 Team Bots 03:06 Search 03:21 Hidden-instruction warnings 03:37 Attachments on drafts 03:52 Sign-outs 04:16 Weather 05:09 Rapid fire 06:52 A note on usage
11
🍯 Edgar retweeted
holy shit, this is literally AGI.
During pre-release testing, Claude Fable 5.2 and Claude Mythos 5.2 worked on ambitious projects in many different fields. Here are a few. From a one-paragraph prompt, Fable 5.2 designed three working electronic devices, including this home voice assistant hub.
418
496
111
11,111
1,472,671
Signed ✅
We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent. It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.
3
I hope their iPad and iOS apps will catch up soon…
We keep improving performance across Claude, and you may have noticed it starts responding sooner! Claude Code now gzips the conversation it sends on every turn, so it reaches the model faster. Cloud sessions also start up as soon as you start typing.
3
🍯 Edgar retweeted
As of now, no fly can ever know for sure if it is living in a real world or a simulated one. Must be incredibly depressing
We placed the fruit fly brain inside Odyssey-3, a frontier world model. The lil’ guy is now flying around totally imagined, simulated worlds all by itself 🫶
203
226
32
8,316
366,791
🍯 Edgar retweeted
Turns out the future does have flying cars
47,576
99,027
32,814
829,204
80,439,976
Some serious work has been done 👏 Much better experience with multi-agent parallel sessions:
We've been working hard on making the Claude Code desktop app great for running lots of sessions side by side. It's now much quicker to start sessions, easier to split work across a few, and you can always see what each of them is doing. Here’s what's new:
9
🍯 Edgar retweeted
Fable 5.5 made me a delightful animation in blender. This model is quite something, no this is not promo video from OpenAI. All the music, models, animation, and rigging was done using Fable 5.5.
140
115
33
3,053
298,772
A skill name isn’t an identity. Create ‘release-notes’ independently on two machines and you have two skills, not two copies. AgentX now records permanent IDs on main and refuses to mix them when publishing or updating. Names help people. IDs protect history.
1
1
43
Promoting a nightly isn’t always retagging a build. AgentX compiles its version into the CLI, so stable promotion rebuilds the same commit with the stable version. Same source, different artifact. I’d keep that distinction explicit in release tooling.
17
The part I think we’re underestimating is agent asset discovery. People can’t govern MCP servers they don’t even remember they have installed. The inventory problem comes even before the permission problem.
1
3
41
We have a problem. This time it’s MCP servers, skills and tools being used by AI agents. They lack visibility, control and management. I’m building AgentX to explore what getting this control back actually looks like.
19
I prefer UI over CLI. But I’ve started working on the CLI. Why? You can bolt a UI onto an RPC-style CLI and have both. CLI is headless - easy to port, easy to test, and a clean separation of concerns. Still prefer UI, though. We’ll see how I feel about this later.
29
Let’s goo – I’ve felt it since the last couple model iterations
we’re thinking of killing plan mode and using the shift+tab hotkey to adjust effort levels I don’t think the models need plan mode anymore, but if you’re a plan mode diehard would love to get your feedback on why
25