Today, I completed Day 1 of EFD TextFix, a new project under EfsunDev.
The goal is to build a lightweight Chrome extension that helps users correct spelling, grammar, and punctuation, improve readability, and rewrite text without losing its original meaning.
Instead of jumping straight into implementation, I focused on making the right decisions first.
What I accomplished
Defined the MVP scope and core features.
Documented the technical requirements and project architecture.
Established an initial API budget and usage limits.
Created a dedicated AI model evaluation plan.
Organized project documentation and status tracking.
Added a proprietary license and copyright notice.
The planned architecture
The current proposal is intentionally simple:
Chrome Extension → Cloudflare Workers → OpenAI API
The extension will handle the user interface, while a minimal backend will validate requests, enforce usage limits, and keep API credentials out of the client-side code.
The initial experiment has a planned API budget of $5 and a provisional global limit of 300 requests per month.
These are project constraints, not measured results. The architecture and model still need to be validated.
Why I haven't started coding yet
The initial AI model candidate is GPT-6 Luna, but it has not been approved for production.
Before implementation, I want to evaluate correction accuracy, rewriting quality, meaning preservation, response time, and actual API costs using both English and Russian text.
The objective is to avoid unnecessary complexity and build only what the experiment requires.
What's next?
Day 2 will focus on preparing test cases and evaluating the AI model before making an implementation decision.
EFD TextFix is a small, one-month product experiment — an opportunity to learn, ship something useful, and strengthen EfsunDev's development process.
Build incrementally. Validate early. Keep it simple.
This is Day 1. More updates to come.
Top comments (0)