How many credentials are sitting in your AI prompts right now?

Your engineers are shipping context to Claude, GPT, etc. System prompts, tool definitions, code, knowledge bases. Some of that probably has secrets in it. API keys, passwords, tokens. Stuff that should never leave your firewall.

Most teams have no idea. We measure this and it's the fastest way to get an engineering leader's attention.

What secrets are you accidentally sharing with your AI providers?