Overview

Relevant source files

• .github/PULL_REQUEST_TEMPLATE.md
• PROMPTS.md
• README.md
• SELF-HOSTING.md
• packages/prompts.chat/README.md
• prompts.config.ts
• prompts.csv
• scripts/generate-contributors.sh
• scripts/rebuild-history.sh
• src/app/prompts.csv/route.ts
• src/components/admin/import-prompts.tsx
• src/components/layout/header.tsx
• src/lib/config/index.ts
• src/lib/i18n/config.ts

Purpose and Scope

This document provides a comprehensive overview of prompts.chat, an open-source prompt library platform that enables users to collect, organize, and share AI prompts across multiple AI models (ChatGPT, Claude, Gemini, Llama, Mistral, and others). This overview covers the platform's architecture, dual persistence model, core integrations, and configuration system.

For detailed information about specific subsystems, refer to:

• Installation and local setup: Getting Started
• Dual persistence implementation: CSV Sync System & Contributor Attribution
• Application structure and routing: Application Structure
• Configuration options: Configuration System
• Database models: Database Schema
• Prompt management: Prompt System
• Authentication: Authentication & Authorization
• Multi-language support: Internationalization

---

What is prompts.chat

prompts.chat is a Next.js-based web application that serves as a curated prompt library with the following characteristics:

Characteristic	Description
License	CC0 1.0 Universal (Public Domain)
Primary Language	TypeScript with Next.js 14 (App Router)
Database	PostgreSQL with Prisma ORM
Authentication	NextAuth v5 (GitHub, Google, Apple, Azure AD, credentials)
Internationalization	17 languages with RTL support
Deployment	Docker, Vercel, or self-hosted
API Protocols	REST API + MCP (Model Context Protocol)

The platform supports five prompt types:

1. TEXT - Simple text prompts
2. STRUCTURED - JSON/YAML prompts with schema validation
3. SKILL - Multi-file agent skills (.skill format)
4. TASTE - Coding style definitions
5. MEDIA - Image/video/audio-enhanced prompts

Sources: README.md1-67 prompts.config.ts1-92 src/lib/config/index.ts1-227

---

Architecture Overview

Key architectural decisions:

1. App Router: Uses Next.js 14 App Router with server components for improved performance
2. Dual Persistence: Database for live data + CSV for git-tracked history
3. Server-Side Rendering: Pages rendered on server with client-side hydration
4. API-First: All data operations through REST or MCP APIs
5. Configuration-Driven: Single prompts.config.ts file controls all customization

Sources: README.md1-242 src/components/layout/header.tsx1-673 src/lib/config/index.ts1-227

---

Dual Persistence Architecture

The platform implements a bidirectional sync between PostgreSQL (live data) and prompts.csv (git history) to achieve multiple goals:

• Git attribution: Each prompt gets its own commit with proper author attribution
• Portability: CSV format allows easy migration and backup
• Community transparency: All prompts visible in git history
• Database performance: Fast queries, filtering, and search

Export Pipeline

Export process:

1. Filtering src/app/prompts.csv/route.ts28-33: Excludes private, unlisted, and deleted prompts
2. Contributor resolution src/app/prompts.csv/route.ts17-24:
   • GitHub OAuth users: use githubUsername
   • Imported contributors: use username (if email ends with @unclaimed.prompts.chat)
   • Others: use email
3. CSV formatting src/app/prompts.csv/route.ts4-15: Escapes quotes, commas, newlines
4. Git commit creation scripts/generate-contributors.sh133-157: Each prompt gets individual commit with proper author

Import Pipeline

Import process:

1. Admin-only access: Requires role: ADMIN to trigger import
2. Duplicate detection: Skips prompts with matching title
3. User resolution: Matches contributor field to existing users or creates unclaimed accounts
4. Attribution: Sets isImported: true flag to distinguish imported prompts

Sources: src/app/prompts.csv/route.ts1-94 scripts/generate-contributors.sh1-431 src/components/admin/import-prompts.tsx1-222

---

Core Integrations

MCP Protocol Integration

The platform implements the Model Context Protocol for AI agent integration:

Component	Path	Description
MCP Server	src/app/api/mcp	Handles MCP protocol requests
Tools	10 registered tools	search_prompts, get_prompt, save_prompt, improve_prompt, etc.
Authentication	API key-based	Uses PROMPTS_API_KEY header
Capabilities	prompts/list, prompts/get	List and retrieve prompts

Configuration example for VS Code/Cursor:

Sources: README.md138-163 packages/prompts.chat/README.md1-61

CLI Tool

Command-line interface for browsing and executing prompts:

CLI capabilities:

• Interactive prompt browser with search
• Variable filling dialog
• Direct prompt execution (opens in ChatGPT, Claude, etc.)
• Clipboard copy with variable substitution

Sources: packages/prompts.chat/README.md27-61

Browser Extensions

Platform-specific extensions for Chrome and Firefox:

Browser	URL	Integration
Chrome	branding.chromeExtensionUrl	Context menu + sidebar
Firefox	Mozilla Addons	Firefox-specific package

Extensions provide:

• Right-click context menu for prompt insertion
• Sidebar with prompt search
• Variable filling UI
• Direct execution in AI chat interfaces

Sources: src/components/layout/header.tsx84-556 prompts.config.ts16-18

---

Configuration System

The platform uses a single configuration file (prompts.config.ts) with environment variable overrides:

Configuration Sections

Section	Purpose	Key Fields
branding	Visual identity	name, logo, logoDark, description
theme	Design system	radius, variant, density, colors
auth	Authentication	providers[], allowRegistration
i18n	Languages	locales[], defaultLocale
features	Feature flags	privatePrompts, aiSearch, mcp, comments
homepage	Landing page	useCloneBranding, sponsors, achievements

Environment Variable Overrides

All configuration can be overridden at runtime using PCHAT_* environment variables:

This allows Docker deployments to customize without rebuilding:

Sources: prompts.config.ts1-92 src/lib/config/index.ts79-227

---

Internationalization System

Platform supports 17 languages with full RTL support:

Language Group	Locales	Special Support
Western	en, es, pt, fr, de, it, nl	LTR with Latin script
Asian	zh, ja, ko	LTR with CJK fonts
Middle Eastern	ar, fa, he	RTL with automatic dir attribute
Other	tr, az, ru, el	LTR with locale-specific fonts

I18n Architecture

RTL language handling src/lib/i18n/config.ts8-14:

• Automatic dir="rtl" attribute on <html> element
• Font switching (e.g., Noto Sans Arabic for Arabic)
• Layout mirroring for RTL languages

Translation keys organized by domain:

• nav.* - Navigation labels
• prompts.* - Prompt management
• admin.* - Admin interface
• errors.* - Error messages
• feed.* - Feed and discovery

Sources: src/lib/i18n/config.ts1-25 prompts.config.ts44-47

---

Deployment Options

The platform supports three deployment models:

1. Docker (Self-Hosted)

Quick start:

Features:

• Embedded PostgreSQL (managed by supervisord)
• Automatic migrations on startup
• Volume persistence for /data
• Multi-architecture support (amd64, arm64)

2. Vercel (Serverless)

Deployment:

Features:

• Automatic serverless functions
• Edge runtime support
• External PostgreSQL required
• Instant scaling

3. Self-Hosted (Manual)

Setup:

Requirements:

• Node.js 18+
• PostgreSQL database
• OAuth credentials (GitHub/Google/etc.)

Sources: README.md100-129 SELF-HOSTING.md1-206

---

Key File Paths

Path	Purpose
prompts.csv	Git-tracked prompt storage (3898.31 importance)
PROMPTS.md	Human-readable prompt catalog (2448.53 importance)
prompts.config.ts	Main configuration file (24.43 importance)
scripts/generate-contributors.sh	CSV-to-git sync script (19.44 importance)
src/app/prompts.csv/route.ts	CSV export API endpoint (5.96 importance)
src/lib/config/index.ts	Configuration loader (8.14 importance)
src/lib/i18n/config.ts	I18n configuration (4.61 importance)
src/components/layout/header.tsx	Main navigation (29.58 importance)

Sources: prompts.csv1-10 PROMPTS.md1-100 prompts.config.ts1-92 scripts/generate-contributors.sh1-431

---

This overview provides the foundation for understanding prompts.chat's architecture. For detailed information about specific subsystems, refer to the section-specific wiki pages listed at the beginning of this document.