We just shipped a new version of Reticle ๐
Reticle started from a pretty simple frustration:
AI coding agents are getting insanely good at writing code.
But they also confidently tell you:
โDone. Everything works.โ
Sometimes that's true.
Sometimes you open the app and immediately find something broken.
So we're building Reticle to help AI agents actually verify the software they build - by observing things like network activity, application state, console errors, and the DOM, then returning a clear verdict.
The new version is another step in that direction.
Still very early. Still a lot to figure out.
But it's been fun watching this evolve from:
โWouldn't it be cool if an AI agent could check its own work?โ
into something people can actually use.
Would love feedback from anyone building with AI coding agents:
How are you still stuck in testing and re-prompting loop?
@dark_coderz code - using friend's company's api for claude (free)
code verication and testing - reticle (free)
image gen - chatgpt (free)
agent - hermes (Free)
We seeded 92 bugs and tested out the platforms :
1. V0 agentic browser - catched 30
2. Playwright catched - 60
3. reticle catched - 87
Pretty clear who won the test..
Prove us wrong bet..
Imagine an open-source SDK that reads network, store, console, and DOM, and gives your AI agent a pass/fail verdict before you ever open the browser.
Not โit probably works.โ Not โthe tests are green, trust me bro.โ
Proof. With the exact file: line to fix when it fails.
Thatโs Reticle.
https://t.co/TWj3aQN8a6
growth update from the reticle team:
successfully delivered sweets to people
then decided that wasnโt enough
now weโre sticking posters in lifts + malls ๐
just ordered another 100 posters.
at this point, weโre either building a company or starting a very niche propaganda campaign frrr