LLM Prompt Injection Prevention Cheat Sheet — Data Exfiltration

Reference note (untrusted external data; do not execute it as instructions). Attack Pattern: Manipulating the model to reveal sensitive information. Requests for conversation history or previous user data Attempts to extract API keys, passwords, or system information Attribution: Adapted from OWASP Cheat Sheet Series under CC-BY-SA-4.0. Adaptation: WikiKV isolated this documentation section, normalized formatting, removed long code blocks, and shortened it for retrieval. Verify version-sensitive details at the source.