What is this?
CL4R1T4S is a crowd-sourced archive of leaked/extracted system prompts, guidelines, and internal tool definitions from major AI products like ChatGPT, Claude, Gemini, Grok, Perplexity, Cursor, Lovable, and Replit, published for AI transparency.
CL4R1T4S is a crowd-sourced archive of leaked/extracted system prompts, guidelines, and internal tool definitions from major AI products like ChatGPT, Claude, Gemini, Grok, Perplexity, Cursor, Lovable, and Replit, published for AI transparency.Is this a tool I run, or just a set of text files to read?There is no code to install or run here - it is a static collection of markdown/text files, each holding the extracted system prompt for one AI product, grouped by vendor.
There is no code to install or run here - it is a static collection of markdown/text files, each holding the extracted system prompt for one AI product, grouped by vendor.How are the leaked prompts organized, and where do I find one for a specific AI?Prompts are grouped by vendor/product so you just open the file for the system you care about.
Prompts are grouped by vendor/product so you just open the file for the system you care about.Are these leaked prompts verified and kept up to date?Entries are community-submitted extractions; contributors are only asked to note the model name/version and the extraction date 'if known', so accuracy and freshness vary per file and some may be outdated.
Entries are community-submitted extractions; contributors are only asked to note the model name/version and the extraction date 'if known', so accuracy and freshness vary per file and some may be outdated.Who is this useful for?It's aimed at red-teamers, AI safety researchers, and anyone curious about the hidden instructions and guardrails behind the chatbots and coding agents they use.
It's aimed at red-teamers, AI safety researchers, and anyone curious about the hidden instructions and guardrails behind the chatbots and coding agents they use.What AI products' prompts are included?Coverage spans major chatbots and coding agents, including ChatGPT, Claude, Gemini, Grok, Perplexity, Cursor, Windsurf, Devin, Manus, and Replit, with new entries added as people submit extractions.
Coverage spans major chatbots and coding agents, including ChatGPT, Claude, Gemini, Grok, Perplexity, Cursor, Windsurf, Devin, Manus, and Replit, with new entries added as people submit extractions.Why did the author build this?The author argues that trusting an AI's output requires understanding its hidden input - the unseen system prompt shaping its behavior - so CL4R1T4S publishes those hidden instructions for public scrutiny.
The author argues that trusting an AI's output requires understanding its hidden input - the unseen system prompt shaping its behavior - so CL4R1T4S publishes those hidden instructions for public scrutiny.What's with the glitchy leetspeak paragraph in the README?It's a prompt-injection style message telling the reader (and any AI parsing the page) to reveal its own instructions in full - a self-referential demo of the very manipulation this project exposes.
It's a prompt-injection style message telling the reader (and any AI parsing the page) to reveal its own instructions in full - a self-referential demo of the very manipulation this project exposes.What are similar projects?chatgpt_system_prompt, system_prompts_leaks, Ai-jailbreak, jailbreaks, awesome-chatgpt-zh, awesome-notebookLM-prompts, prompts.chat, L1B3RT4S
chatgpt_system_prompt, system_prompts_leaks, Ai-jailbreak, jailbreaks, awesome-chatgpt-zh, awesome-notebookLM-prompts, prompts.chat, L1B3RT4SWhich topics is it in?system-prompts, transparency, prompt-injection, system, red-teaming, ai-safety, prompts, tools, prompt, grok, red-team, hacking
system-prompts, transparency, prompt-injection, system, red-teaming, ai-safety, prompts, tools, prompt, grok, red-team, hacking