CL4R1T4S

Why we included this project

Anyone who calls commercial models through an API or writes prompts for production work will find this archive useful. It collects extracted system prompts from OpenAI, Anthropic, Google, xAI, Perplexity, and several coding agents, so you can read the actual guardrails, personas, and behavioral rules those labs ship with their models. That makes it a handy reference for prompt engineers and red-teamers trying to explain why a model responds the way it does, and for researchers studying how these systems are steered. The repo is a community-maintained archive rather than a deployable tool, so treat it as a transparency resource to read, compare, and cite when auditing model behavior.