system-prompts-and-models-of-ai-tools

Why we included this project

This repo collects the system prompts and internal instructions behind a wide range of AI products, from coding assistants like Cursor, Claude Code, and Devin to general tools such as Perplexity, NotionAI, and v0. It's a community-maintained archive, best treated as a reference and study resource rather than something you run directly. If you build or evaluate AI applications, reading how established products frame their behavior and set guardrails around tool use is a practical way to learn prompt design. You can compare how different teams structure their system messages and spot recurring patterns for safety and formatting, then borrow ideas when drafting your own prompts. The collection spans many product categories, so it's especially useful for researchers and prompt engineers, and developers writing their own prompts will find plenty to borrow from.