title	MyPromptHive - Project README
version	4.0.1
last_updated	2026-04-23

MyPromptHive is a self-hosted, local-first Prompt Management System designed for prompt engineers and AI enthusiasts. It allows you to organize, version, and text your LLM prompts in a beautiful, structured interface.

🧪 Experiment Disclaimer: This project is an experiment on the limits of "vibe coding". It was completely vibe coded, including all documentation, testing, and infrastructure configuration. No manual coding was performed.

📜 Changelog

See CHANGELOG.md for the full history of changes.

🚀 Comprehensive Feature Overview

📝 Core Prompt Management

• Rich Editor: Syntax highlighting for variables and markdown support.
• Dynamic Variables: Define variables like {{topic}} or [[style]] and auto-generate testing forms.
• Version Control: Automatically track every change.
• Visual Diff: Compare any two versions side-by-side to see exactly what changed (additions/deletions).
• Private Prompts: Mark drafts or sensitive prompts as private so only you can see them.
• Metadata: Add tags, descriptions, source URLs, and attachments to your prompts.

🤖 Agent Skills

• Agent Skill Management: Store and manage AI Agent Skills alongside your prompts with dedicated CRUD, badges (🤖 Skill / 📝 Prompt), and detail views.
• Skill Linking: Link agent skills to prompts for context-aware clipboard actions. Linked skills include descriptions in the prompt detail view.
• Group Skill Import: Import multiple agent skills at once by providing a list of GitHub repository URLs. Skills are automatically organized into a timestamped collection.
• Advanced Copy: Copy prompts with configurable context blocks — toggle Agents, Agent Skills, Source-of-Truth Policy, Persistence Check, and User Handoff sections before copying.

🗂️ Organization & Discovery

• Nested Collections: Organize prompts into infinite folders and sub-folders. Supports Drag & Drop for easy reorganization.
• Advanced Search: Filter by tags, content, or creator.
• Command Palette: Navigate the entire app without leaving the keyboard (Cmd+K).
• Workflows (BETA): Chain multiple prompts together to create complex automation sequences. (Note: This feature is currently in BETA and untested)
• Favorites & History: Quickly access your most used or recently viewed prompts.

🔄 Data & Integration

• Unified Import: Seamlessly import data from backups or exports with two-phase agent skill link resolution.
• AI Web Scraper: Built-in guide and prompt to let AI agents scrape prompts from any website for you.
• Export Options: Download your entire library as JSON or individual prompts as Markdown. Export includes full agent skill metadata and cross-references.
• Automated Backups: Configure daily, weekly, or monthly automatic backups to a local path.

⚙️ System & Administration

• Local First: 100% self-hosted. All data lives in a local SQLite database on your machine.
• Internationalization (i18n): Fully localized in English, German, French, Spanish, Italian, Dutch, and Swedish.
• User Management: Admin dashboard to manage users and toggle registration.
• Customization: Toggle between Dark and Light modes.
• Robust Security: Local authentication with strictly typed schema validation.

🎥 Demos

Demo videos can be found on the following YouTube Channel.

🛠️ Technology Stack

• Framework: Next.js 16+ (App Router)
• UI Library: React 19
• Database: SQLite with Prisma ORM 6+
• Styling: Tailwind CSS + Custom Design System
• Authentication: NextAuth.js (Local Credentials)
• Validation: Zod schema validation
• Testing: Vitest (Unit) & Playwright (E2E)

🏁 Getting Started

Prerequisites

• Node.js 18+
• npm

Installation

1. Clone the repository

   git clone https://github.com/yourusername/myprompthive.git
   cd myprompthive

2. Install dependencies

   npm install
   # and
   npm install -g @mermaid-js/mermaid-cli # For doc generation

3. Running the Application

   This project supports two environments: Development and Production.

   Option A: Development / Test

   • Port: 3000
   • Database: data/dev.db
   • Use for: Coding, testing, and debugging.
   • Run it this way if you: Are actively modifying code, debugging new features, or want to see changes update in real-time.
   • Pros: Hot Reloading, Detailed Error Overlays, Easy Debugging.
   • Cons: Slower performance, less strict than production environment.

   # 1. Initialize Database
   npx prisma db push
   
   # 2. Start Dev Server
   npm run dev

   Open http://localhost:3000.

   Option B: Production

   • Port: 3100
   • Database: data/prod.db (Separate stable database)
   • Use for: Daily usage.
   • Run it this way if you: Want to use the stable application locally with maximum performance and no container overhead, but don't plan to modify the code.
   • Pros: Maximum Performance, Persistent Data (Separate DB), Stable.
   • Cons: No Hot Reloading (requires restart for changes), harder to debug errors.

   # 1. Build the Application
   npm run build
   
   # 2. Initialize/Update Production Database
   npm run db:push:prod
   
   # ALTERNATIVE: Start with EMPTY database (Wipes existing data!)
   # npm run db:reset:prod
   
   # 3. Start Production Server
   npm run start

   Open http://localhost:3100.

   Option C: Docker (Production)

   Run the production build in an isolated container.

   • Run it this way if you: Want a completely isolated, reproducible environment that mimics a real server deployment, or if you want to run the app in the background without worrying about local Node.js version conflicts.
   • Pros: Complete Isolation, Consistent Environment, Easy Deployment/Migration.
   • Cons: Requires Docker, slightly more complex setup (volume mounting), harder to access internal files directly.

   # 1. Build the Image
   docker build -t myprompthive-production .
   
   # 2. Prepare Data Directory
   # Create a 'data' folder in your project root to persist the database.
   # The container will look for /app/data/prod.db
   New-Item -ItemType Directory -Force -Path data
   
   # 3. Run Container
   # Maps local 'data' folder to container storage
   # Uses docker.env for default configuration (Update this file or use .env.production for real secrets)
   docker run -d -p 3100:3000 --name prompthive `
     -v ${PWD}/data:/app/data `
     -v ${PWD}/public/uploads:/app/public/uploads `
     --env-file docker.env `
     myprompthive-production

   Open http://localhost:3100.

⚙️ Configuration

The project uses environment variables for configuration.

• docker.env: Included in the repo with default values for easy testing.
• .env.production: (Ignored by Git) Create this file for secure production deployments to override secrets.
• admin.properties: Contains the password to grant yourself admin rights.

Note on Ignored Files:

• data/*.db: SQLite databases are ignored to prevent sharing your personal data. They are automatically initialized when you run the setup commands.
• public/uploads/*: User uploads are ignored. The folder structure is preserved.

📂 Project Structure

.
├── actions/              # Modular Server Actions (Prompt, Collection, Skills, etc.)
├── app/                  # Next.js App Router (Pages & API)
│   ├── (auth)/           # Login/Register Routes
│   ├── (dashboard)/      # Protected App Routes (Prompts, Skills, Collections)
│   └── api/              # REST API Endpoints
├── components/           # React Components
├── docs/                 # Documentation (Single Source of Truth)
│   ├── technical/        # Architecture & Deep Dives
│   └── diagrams/         # Database Models & Diagrams
├── lib/                  # Utilities (Auth, Validation, etc.)
├── prisma/               # Database Schema
├── services/             # Business Logic Layer (Import, Export, CRUD)
└── public/               # Static Assets

🧪 Testing

Run the test suite (Vitest + Playwright):

npm run test

📖 Documentation

For deep dives, check the docs/ folder:

• Technical Architecture
• Data Models

🔗 Links

• GitHub Repository

🗺️ Roadmap

• Agent Skills Management: Full CRUD for AI Agent Skills with linking, group import, and export. (Shipped in v4.0.0)
• Advanced Copy with Context Blocks: Configurable multi-section clipboard with SOT policy, persistence checks, and user handoff. (Shipped in v4.0.1)
• Skill Marketplace: Browse and install community-contributed agent skills from a public registry.
• Prompt Templates: Create reusable prompt templates with predefined variable slots for team standardization.
• API Access: REST API for external tools and CI/CD pipelines to interact with the prompt library.

Why did I make it?

I built MyPrompthive because I needed a better way to manage my growing collection of prompts. I wanted something that could store prompts efficiently on my own infrastructure, without being locked into any particular AI model or provider. Most existing solutions either lived in the cloud (where I had less control), were tightly coupled to specific APIs, or lacked essential features like proper version history, a flexible folder structure for organization, variable substitution, and attachment support. But equally important was security: I needed a system that could completely decouple prompt management from API credentials. With kids in the house who are naturally curious about AI, I wanted to prevent scenarios where they could accidentally rack up tens of euros in token costs by having access to my API keys. Promthive is that solution: a self-hosted, model-agnostic prompt library that puts organization, security, and cost control first.