Guys, Hear me out.
what if, there is browser which has terminal built-in.
in there we can run AI agents which can easily access website which is open (with our consent) and do testing and etc..,
Come up with more use cases. because I'm almost done building it.
Typo'd URLs shouldn't end in a dead 404.
I built IntelRoute, open-source URL recovery for Next.js:
/abuot → /about
/sign-in → /login
/blgo/helo-world → /blog/hello-world
/wp-admin → 404 (left alone)
No AI, no API keys.
@KyleWang1087922 Real slugs always win: an exact match is checked before any fuzzy matching, so /post-1 and /post-2 never get confused.
For a typo like /post-3, it only redirects if one candidate leads the next by a clear margin. Slugs one character apart never do, so it suggests both.
Setup is 2 commands:
npm i @intelroute/next
npx intelroute discover
It scans your routes, reads page metadata and writes a manifest. Add the proxy and you're done.
Core engine: zero deps, runs on Node, edge and browser. MIT.
https://t.co/ynQdU2yNtQ
Stars welcome ⭐
How it decides:
• Exact or normalized match → redirect
• Confident typo match → 307 redirect
• A few plausible pages → 404 + "Did you mean…?"
• Nothing fits → plain 404
When it's unsure, it suggests. It never guesses a redirect.
check out https://t.co/ul8xMGC6Kt
its a tray app which automatically detects which repo your agent in working on and has every source control features.
Developer support includes Python, JavaScript, HTTP multipart uploads, and Base64.
One caveat: image support is experimental and not production-ready yet.
Worth exploring if you're building AI agents or multimodal apps.
Docs: https://t.co/NWvVpXMb95
🧵 AI models are getting better at understanding images. But what if you could make image understanding structured, programmable, and predictable?
TypeSafe introduces image support for System One, powered by Jev.
Here's what makes it interesting. 👇
What can you build with it?
• Visual classification using reference images.
• Compare images and identify differences.
• Analyze multiple camera frames.
• Ask questions about image content.
Images can be included in state, instructions, and criteria.
@swordBluesy Good test. Manual staging commits only what you staged, but with nothing staged, one-click commit runs git add -A and would sweep in the unrelated file. That’s wrong for an agent tool, so I’m fixing it to stage only the agent’s files. Update coming soon.
launching VibecodeGit - better source control for AI coding https://t.co/ul8xMGC6Kt
for windows for now. Mac soon
A simple app that detects your agents workspace and lets you manage changes, commit, pull & push right from your system tray.
Open source
https://t.co/gikaAx8VGc
Claude code desktop's source control sucks.
So I built a source control utility app which automatically detects which directory claude is working.
It has everything VS code's source control has.
Suggest some names below.
launching VibecodeGit - better source control for AI coding https://t.co/ul8xMGC6Kt
for windows for now. Mac soon
A simple app that detects your agents workspace and lets you manage changes, commit, pull & push right from your system tray.
Open source
https://t.co/gikaAx8VGc
Claude code desktop's source control sucks.
So I built a source control utility app which automatically detects which directory claude is working.
It has everything VS code's source control has.
Suggest some names below.
launching VibecodeGit - better source control for AI coding https://t.co/ul8xMGC6Kt
for windows for now. Mac soon
A simple app that detects your agents workspace and lets you manage changes, commit, pull & push right from your system tray.
Open source
https://t.co/gikaAx8VGc