# Awesome AI System Prompts — 32+ Tool Prompts Revealed

> Curated collection of extracted system prompts from 32+ production AI tools including ChatGPT, Claude, Cursor, v0, Manus, Devin, Windsurf, and Perplexity. MIT license, 5,700+ stars.

## Install

Paste the prompt below into your AI tool:

## Quick Use

1. Browse by tool: [github.com/dontriskit/awesome-ai-system-prompts](https://github.com/dontriskit/awesome-ai-system-prompts)
2. Each tool has its own directory with the full extracted system prompt
3. Study the patterns: role definition, tool integration, safety protocols
4. Apply architectural patterns to your own AI agents

---

## Intro

Awesome AI System Prompts is a curated collection of system prompts extracted from 32+ production AI tools, with 5,700+ GitHub stars and MIT license. It includes the actual system prompts powering ChatGPT (GPT-4o, GPT-4.5), Claude (Opus, Sonnet, Claude Code), Cursor, Windsurf, v0, same.new, Manus, Devin, Perplexity, Grok, Notion AI, and 20+ more. Beyond just collecting prompts, the README analyzes architectural patterns — how the best AI tools define roles, structure instructions, integrate tools, handle safety, and maintain consistency.

Best for: AI agent builders who want to learn from how production AI tools are actually architected.
Works with: any LLM — study the patterns and apply to your own agents.
Updated: regularly as new tools and versions are released.

---

## Awesome AI System Prompts — Tools Covered

### Code & UI Generation
- **Vercel v0** — AI UI component generator
- **same.new** — Clone any website with AI
- **Bolt.new** — AI full-stack app builder
- **Loveable** — AI app builder

### General Purpose Agents
- **ChatGPT** (GPT-4o, GPT-4.5) — OpenAI's flagship
- **Claude** (Opus 4.6, Sonnet 4.6) — Anthropic's models
- **Grok** (4.2, 4) — xAI's assistant
- **MetaAI / Llama** — Meta's assistant
- **Manus** — General AI agent

### Coding Assistants
- **Claude Code** — Anthropic's CLI agent
- **Cursor** — AI code editor
- **Windsurf** — Codeium's IDE
- **Augment** — Enterprise coding AI
- **Cline** — VS Code agent
- **Devin** — Autonomous software engineer

### Specialized Tools
- **Perplexity** — AI search engine
- **Notion AI** — Workspace assistant
- **Replit** — Online coding platform
- **Blackbox.ai** — Code search
- **Google Jules** — Google's coding agent

### Architectural Patterns Analyzed

The README identifies 8 core principles shared across top AI tools:

1. **Clear Role Definition** — Every tool starts with explicit identity
2. **Structured Instructions** — Organized, hierarchical rules
3. **Tool Integration** — How tools expose and manage capabilities
4. **Step-by-Step Reasoning** — Enforced thinking patterns
5. **Environment Awareness** — Context about platform, user, time
6. **Domain Expertise** — Specialized knowledge injection
7. **Safety Protocols** — Content filtering, refusal patterns
8. **Interaction Consistency** — Tone, format, behavior standards

### FAQ

**Q: What is Awesome AI System Prompts?**
A: A collection of extracted system prompts from 32+ production AI tools (ChatGPT, Claude, Cursor, Devin, etc.) with analysis of architectural patterns.

**Q: Is it free?**
A: Yes, MIT license.

**Q: How are prompts extracted?**
A: Through various methods including prompt injection, source code analysis, and community contributions.

---

## Source & Thanks

> Maintained by [dontriskit](https://github.com/dontriskit). Licensed under MIT.
>
> [awesome-ai-system-prompts](https://github.com/dontriskit/awesome-ai-system-prompts) — ⭐ 5,700+

---

<!-- ZH -->

## 快速使用

浏览：[github.com/dontriskit/awesome-ai-system-prompts](https://github.com/dontriskit/awesome-ai-system-prompts)

---

## 简介

从 32+ 个生产 AI 工具中提取的系统提示集合，5,700+ Star，MIT 许可证。包括 ChatGPT、Claude、Cursor、Devin、Manus、v0、Perplexity 等的实际系统提示，并分析了 8 个核心架构模式。

---

## 来源与感谢

> Maintained by [dontriskit](https://github.com/dontriskit). Licensed under MIT.
>
> [awesome-ai-system-prompts](https://github.com/dontriskit/awesome-ai-system-prompts) — ⭐ 5,700+

---
Source: https://tokrepo.com/en/workflows/7d697443-cf7f-4e71-8fd7-45431e5ff685
Author: Prompt Lab