The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes documentation with adversarial prompt examples and Base64-encoded injection strings used to demonstrate how systems should be designed to identify and refuse malicious instructions.
[DATA_EXFILTRATION]: The skill provides robust guidelines and implementation patterns for PII redaction and secure logging to prevent sensitive data exposure in production environments.
[SAFE]: Analysis of the skill's workflow and templates confirms a focus on least privilege, human-in-the-loop approvals, and proactive safety monitoring.

building-with-llms