For the last 6 months, I’ve been head-down building a project called ReCode.
But when you build in a silo for half a year, it’s easy to get tunnel vision. I am at a crossroads where I need to step back and ask the community for a brutal reality check.
What is ReCode?
At its core, ReCode is a development ecosystem and AI workbench. My goal was to create a hybrid platform—bridging a SaaS dashboard with a self-hosted, "Bring Your Own Model" setup for developers and teams who cannot legally (or safely) paste proprietary code into public AI endpoints.
Instead of building another generic chat interface(which im really trying NOT to), I focused doing the basic boring tasks and the problem of forgetful AI web tabs interfaces. Of course there are things like Cursor and Claude Code, but i dont use them.
Here is what the engine actually handles right now:
1. The Enterprise Mock Data Factory
Generating realistic, interconnected mock data is usually a massive headache. ReCode parses your SQL DDL, Prisma schema, or TypeScript interfaces and generates production-ready, relational mock data. It automatically maps foreign keys to primary keys across multiple tables and applies chronological/mathematical behavioral rules.
2. AI Code Analysis & Security Auditor
It runs a full-spectrum health report covering:
Security Audits, Complexity, Architecture Review and more.
3. Universal Translation & Refactoring
Migrating logic across languages (e.g., Python to TypeScript, C++ to Rust).
The Reality Check I Need From You
The tooling market right now is heavily dominated by inline IDE AI (Cursor, Windsurf, GitHub Copilot). I know that developers hate breaking their flow to open a browser tab.
So, here is the question I am wrestling with before I write another line of code:
Does a dedicated, privacy-focused workbench solve a friction point you actually hit in your daily workflow? Or is having an AI chat tab right inside your IDE genuinely enough for you?
Would you use a standalone tool for heavy-lifting tasks like generating relational database seeds or running a deep architectural audit?
Would local-model self-hosting actually get you to use this, or is that still not enough?
I’m looking for cold, honest feedback. If I need to pivot this into an IDE extension, or if I should double down on the self-hosted privacy angle, I want to hear it.
Tear it apart in the comments. Thanks for your time!
Top comments (0)