Promptor
A prompt management system for AI power users. Searchable library with tags, favorites, and categories built on Supabase for real-time data with row-level security. PWA offline access via Serwist service workers. Includes an AI optimizer powered by OpenAI that analyzes and suggests prompt improvements. Demonstrates database design, auth flows, service worker caching strategies, and API integration patterns.
My Thoughts
Why I Built This
Working with AI daily, I accumulated hundreds of prompts across different tools. I needed a single place to store, search, tag, and improve them.
Architecture
Supabase handles auth (Google + email), database (PostgreSQL), and row-level security so each user only sees their own prompts. The AI optimizer endpoint sends prompts to OpenAI for analysis and improvement suggestions.
Key Engineering Decisions
- Row-level security policies in Supabase mean the app is multi-tenant with zero backend auth code.
- PWA with service workers via Serwist enables offline prompt access, important for working in low-connectivity environments.
- Tag-based filtering uses Supabase array columns with GIN indexes for fast lookups at scale.