AI Prompts Library & Skills Hub
A production-grade platform serving a curated collection of AI prompts, agent skills, and automated workflows.
1. Overview
The Prompts Library & Skills Hub is a live production platform serving as a centralized repository for AI capabilities. It publicly hosts 39 engineered prompts, 82 agent skills, and 4 automated workflows.
2. Engineering Highlights
- Architecture: Next.js 16 (App Router) with TypeScript, Tailwind CSS, Base UI, NextAuth v5, and Neon Serverless PostgreSQL.
- State & Data Flow: Relational database schema engineered with Drizzle ORM (
prompts,skills,skill_workflows,prompt_collections), featuring custom JSON validation scripts and Server Actions. - Key Challenge & Solution: Managing multi-agent skill taxonomy and workflow orchestration across 82 skills and 4 workflows; solved by architecting an idempotent data import pipeline (
db:import-skills) and resilient JSONB schema mapping.
Screenshots
Prompt Library & Search
Prompt Detail
Multi-step Workflows
3. Context & Problem
As the usage of LLMs and autonomous AI agents scales, managing system prompts, agent skills, and workflow configurations in scattered local files becomes inefficient. There was a strong need for a structured, public-facing hub where users could easily discover, organize, and utilize these instructions.
4. Goal
To build a scalable application to host a growing library of AI prompts and agent skills, and robustly deploy it to a real production environment where users can start integrating these resources immediately.
5. My Role & What I Built
I single-handedly architected and developed the entire application from Phase 1 through Phase 14 (100% completion).
- Full-stack System: Built both the frontend interface (with i18n support) and backend logic.
- Database Architecture: Engineered the database using Neon Postgres and Drizzle ORM to efficiently structure relational data between prompts, skills, and multi-step workflows.
- Production Pipeline: Ensured codebase stability (
tsc, linting, build checks) and successfully deployed to a live production domain.
6. Result & Current State
The project is fully live and operational. It successfully serves 39 live prompts and a comprehensive hub covering 82 distinct agent skills. All 14 planned developmental phases were executed to 100% completion on the master branch.
7. Demo & Code
- Production Live: prompts.chinnakrit.dev/th/prompts
- GitHub (Private Repository): Manchinn/chinnakrit-prompts