The Agent Skills Directory

[SAFE]: The skill instructions and associated reference files do not contain any malicious code, command execution, or data exfiltration patterns. The skill is focused on structured interaction logic and documentation.
[INDIRECT_PROMPT_INJECTION]: While the skill processes user input to adaptively generate interview questions, it implements a robust mitigation strategy. The mandatory 'recap and confirm' protocol requires the agent to summarize all gathered information and obtain explicit user confirmation before generating final recommendations or invoking downstream tools. This prevents the agent from acting on potentially malicious instructions embedded in user responses without oversight.
[COMMAND_EXECUTION]: The skill includes a 'Quick diagnostic' bash snippet (echo "$USER_PROMPT" | head -c 200). This is a benign diagnostic pattern used for conceptualizing decision-making logic and does not pose a threat of privilege escalation or arbitrary code execution.

guided-interview