Looking for ~10 people to try an early build of a prompt-scoring browser extension
Built a browser extension called Prompt Check. It scores your AI prompts (ChatGPT/Claude/Gemini/Perplexity) out of 100 before you send them and flags specific things you left out, like no stated output format or no instruction for what the model should do when it isn't sure.
Not on the Chrome/Firefox/Edge stores yet (all pending review), so I'm looking for a handful of people willing to sideload a test build and just use it normally for a few days. About 2 minutes to install.
What I'm actually after: people who use ChatGPT/Claude regularly for real work, not people who install it once and forget. Honest "I turned it off after a day" is more useful to me than polite silence.
Runs entirely locally, no account, no API key, no telemetry. The build fails if any network-call code gets added, so that's enforced, not just a claim in a README.
Install: https://github.com/blueyferg/prompt-check/releases/tag/pilot-v0.1
Comment or reply here if you want in, happy to look at your product too.
The local-only part is interesting to me, especially for a browser extension. I've found that users can be surprisingly cautious about what an extension can access, even when the architecture is specifically designed not to send their data elsewhere.
I think making that architecture visible and understandable might be almost as important as implementing it. “Runs entirely locally” is clear to us as developers, but I'm curious whether non-technical users actually understand what that means for their data.
Have you had any feedback yet specifically about the local/no-telemetry aspect?
Looks very interessting, I'll test it for sure. Is it fine to just open Github issues if I find something not working, or improvement suggestions?