Divyendra Singh

[Divyendra Singh @Divten7]

Been working on a new feature. Demo soon!

[Divyendra Singh @Divten7]

Added spatial context to my desktop agent.

Now I can just point at something on my screen and ask it to act on it.

No need to explain what I’m referring to. It knows exactly where I pointed.

[Divyendra Singh @Divten7]

Added spatial context to my desktop agent.

Now I can just point at something on my screen and ask it to act on it.

No need to explain what I’m referring to. It knows exactly where I pointed.

[Divyendra Singh @Divten7]

@thsottiaux Yeah, it looks that way

[Divyendra Singh @Divten7]

They’re really not letting me rest much 😭

[Claude @claudeai]

We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.

[Divyendra Singh @Divten7]

Made a small website for the desktop agent I’ve been sharing lately.

Still building it, but here’s the first look:
https://t.co/gWGzFKuiQd

[Divyendra Singh @Divten7]

Made a small website for the desktop agent I’ve been sharing lately.

Still building it, but here’s the first look:
https://t.co/gWGzFKuiQd

[Divyendra Singh @Divten7]

@GarvitXalt lfg!!

[Divyendra Singh @Divten7]

@yash_darbar16 The MCP gives Claude visual context screenshots plus the exact element/coordinates we’re pointing at so it can understand the reference and act on it accurately.

[Divyendra Singh @Divten7]

One thing I kept running into with coding agents was how hard it is to give them visual context.

I ended up building an MCP server that lets the agent see what I'm looking at. I can point at something on the screen, talk through it naturally, and it understands exactly what I'm referring to.

It's a surprisingly small change, but it makes working with a coding agent feel much more natural.

[Farza 🇵🇰🇺🇸 @FarzaTV]

Today we're shipping spatial context.

This lets you draw directly on your screen + pass it as context to the model.

AI can easily analyze screenshots. But, it has no idea where exactly you want it to look. Now you can just show it.

It's very intuitive.

Demo:

[Divyendra Singh @Divten7]

It's kind of wild watching Fable work for 47 minutes straight on a complex task. Feels a lot more hands free than before you can actually focus on something else while it keeps making progress...

[Divyendra Singh @Divten7]

Btw, here's the final result

[Divyendra Singh @Divten7]

@FarzaTV Been building around this too. The biggest thing I've noticed is how much time gets wasted just trying to explain what you're looking at. Once the agent has that context, everything feels a lot smoother.

Demo:

[Divyendra Singh @Divten7]

One thing I kept running into with coding agents was how hard it is to give them visual context.

I ended up building an MCP server that lets the agent see what I'm looking at. I can point at something on the screen, talk through it naturally, and it understands exactly what I'm referring to.

It's a surprisingly small change, but it makes working with a coding agent feel much more natural.

[Divyendra Singh @Divten7]

Been working on a few more things since this demo. Next one coming soon.

[Divyendra Singh @Divten7]

I've been building a desktop agent that can understand intent and complete multi-step tasks across your desktop.

The part I find most exciting is that it works in the background, so you can keep using your computer while it gets things done.

[Divyendra Singh @Divten7]

I've been building a desktop agent that can understand intent and complete multi-step tasks across your desktop.

The part I find most exciting is that it works in the background, so you can keep using your computer while it gets things done.

[Divyendra Singh @Divten7]

@piyushgarg_dev @grok @grok mine?

[Divyendra Singh @Divten7]

@GarvitXalt Why use any of them??