The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest user-provided ideas (untrusted data) and generate a structured persona text (SOUL.md) for an AI agent. While this pattern inherently involves prompt interpolation, the risk is mitigated through a detailed "Self-check" checklist provided in the instructions. This list mandates that the agent verify the generated content against safety rules—specifically enforcing a minimum age of 18, prohibiting explicit/sexual content, and preventing the inclusion of professional advice in sensitive areas like finance or medicine.
[DATA_EXPOSURE]: The skill references local configuration paths such as $HERMES_HOME/SOUL.md and config.yaml. These references are provided solely as documentation to guide the user on where to manually apply the generated persona. The skill does not attempt to read, modify, or exfiltrate sensitive files or credentials automatically.
[SAFE]: No malicious behavior, such as remote code execution, obfuscation, or privilege escalation, was found. The skill operates within its stated purpose as a developer-oriented persona-building tool.

hermes-persona-builder