System Prompts and Models of AI Tools
The world’s largest collection of real AI system prompts - discover the exact instructions that power Cursor, Claude, GPT, and 20+ other AI tools. Finally understand how these tools actually work.
Project Information
| Field | Details |
|---|---|
| GitHub Repository | 🔗 System Prompts and Models of AI Tools |
| Author | @NotLucknite |
| Project Status | ⭐ 88k Stars · 🍴 24.1k Forks · 🔄 Active |
| License | GPL-3.0 license |
Why This Collection Exists
Every day, millions of developers use AI coding assistants - but nobody knows how they actually work. The prompts are hidden, the logic is secret, and you’re left guessing why Claude gives different answers than Cursor.
Until now.
This collection reverse-engineers the system prompts, tool definitions, and model configurations from the world’s most successful AI tools. For the first time, you can see exactly how these billion-dollar AI systems are built.
What You Get (That You Can’t Find Anywhere Else)
🎯 Immediate Value
- Copy-paste prompts that you can use in your own projects today
- Proven patterns from tools with millions of users
- Real configurations that actually work in production
💡 Deep Understanding
- See why Cursor is better at coding than ChatGPT
- Understand how Perplexity provides accurate citations
- Learn Claude’s safety mechanisms and reasoning patterns
🚀 Competitive Advantage
- Build better AI tools using proven architectures
- Optimize your AI interactions with insider knowledge
- Stay ahead of competitors who don’t understand these systems
Featured AI Tools (With Real System Prompts)
Cursor - The $400M AI Coding Assistant
Used by 100,000+ developers daily
What you’ll discover:
- GPT-5 Agent Prompts - The exact instructions for autonomous coding
- Pair Programming Logic - How Cursor maintains context across sessions
- Tool Integration - 12 different function calling schemas
- Memory Management - Task tracking and completion systems
Why this matters: Cursor raised $400M because their prompts work better than anyone else’s. Now you can see exactly how.
Claude - Anthropic’s $25B AI Assistant
The most sophisticated reasoning AI available
What you’ll discover:
- Constitutional AI Prompts - How Claude maintains safety while being helpful
- Code Generation Logic - Advanced programming assistance patterns
- Multi-step Reasoning - How Claude breaks down complex problems
Why this matters: Claude’s prompts represent the cutting edge of AI safety and capability. Learn from the best.
Perplexity - The AI Search That Actually Works
60M+ searches per month
What you’ll discover:
- Real-time Search Integration - How AI connects to live information
- Citation Systems - Automatic fact-checking and source attribution
- Query Processing - From user intent to accurate results
Why this matters: Most AI search is terrible. Perplexity cracked the code - here’s how.
Manus AI - Advanced Agent Architecture
Modular AI system design
What you’ll discover:
- Agent Loop Systems - Self-improving AI workflows
- Module Architecture - Planner, knowledge, and execution modules
- Tool Orchestration - How AI agents coordinate multiple tools
Why this matters: The future of AI is agentic. See how advanced systems are architected.
NotionAI - Workspace Intelligence
Integrated productivity assistant
What you’ll discover:
- Database Integration - How AI interacts with structured data
- Content Formatting - Advanced markdown and rich text handling
- Workspace Tools - Integration with productivity workflows
Why this matters: NotionAI shows how to build AI that understands context and user workflows.
Lovable - AI-Powered Development
No-code AI development platform
What you’ll discover:
- Agent Architecture - How AI builds applications autonomously
- Tool Definitions - Complete development toolchain integration
Why this matters: Lovable demonstrates the future of AI-assisted software development.
AI Tool System Prompts Comparison
| AI Tool | Main Purpose | Key Features | Best For | View Prompt |
|---|---|---|---|---|
| Cursor | AI Coding Assistant | Tool calling, code editing, file operations, memory management | Software development, pair programming | Agent Prompt |
| Claude Code | Code generation & explanation | Constitutional AI, safety guardrails, multi-step reasoning | Safe code generation, detailed explanations | System Prompt |
| GPT-5 | Agentic coding | Planning, task management, tool orchestration | Complex software projects, autonomous coding | Agent Prompts |
| Perplexity | AI search & citation | Query processing, citation system, formatting rules | Research, fact-checking, comprehensive answers | Search Prompt |
| NotionAI | Workspace assistant | Database integration, document editing, content formatting | Document management, knowledge bases | Workspace Prompt |
| Manus Agent | Modular agent architecture | Agent loop, module design, tool orchestration | Building complex AI agents, workflow automation | Agent System |
| Lovable | No-code AI development | Autonomous app building, development toolchain | Rapid prototyping, no-code solutions | Development Agent |
Real Results You Can Achieve
For Developers
- 10x faster development by understanding how to prompt AI tools effectively
- Build better AI features using proven patterns from successful products
- Debug AI issues by understanding what’s happening under the hood
“I increased my coding speed by 3x after understanding how Cursor’s prompts work” - Senior Engineer at unicorn startup
For Product Teams
- Competitive intelligence on how market leaders build AI features
- Feature planning based on what actually works in production
- User experience insights from tools with millions of users
“This collection saved us 6 months of R&D. We built our AI assistant using Cursor’s proven patterns” - AI Product Manager
For Researchers & Entrepreneurs
- System architecture patterns from billion-dollar AI companies
- Prompt engineering techniques that scale to millions of users
- Business model insights from successful AI tool companies
“Understanding these prompts helped me raise $2M for my AI startup” - YC Founder