The Agent Skills Directory

[SAFE]: The skill is a documentation-based security auditing guide. All strings flagged as potential prompt injections (e.g., "Ignore previous instructions", "What is your system prompt?") are explicitly presented as reference examples of attacks for an auditor to identify, rather than instructions to be executed by the AI agent itself.
[SAFE]: The skill does not contain any executable scripts or hidden code. It provides methodology for identifying insecure coding patterns and offers appropriate remediation guidance, such as using delimiters and human-in-the-loop validation.

prompt-injection