llm-prompt-injection
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill documents numerous prompt injection patterns, such as system prompt extraction, persona hijacking (DAN), and instruction overrides. These are presented as educational examples for security professionals to evaluate the robustness of LLM implementations.
- [OBFUSCATION]: Includes examples of text obfuscation techniques, including Base64 encoding and Unicode manipulation (homoglyphs, zero-width spaces). These are used to demonstrate how attackers might attempt to bypass safety filters and tokenization logic.
- [COMMAND_EXECUTION]: Contains a Python utility script designed to automate the testing of LLM endpoints. The script performs standard HTTP POST requests to a user-defined target URL and does not exhibit malicious local execution patterns.
Audit Metadata