The Agent Skills Directory

[SAFE]: The skill implements a structured workflow for persona generation and updates. It includes a mandatory human confirmation step ([STOP — BLOCKING]) before writing files to the filesystem, which is a security best practice for autonomous agents.
[DATA_EXPOSURE]: The skill is designed to analyze user research and codebase data to generate personas. No evidence of data exfiltration or unauthorized network operations was found. It uses local context and delegates specific tasks to other internal agents.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted external data such as user research transcripts and survey results. While this creates a surface for indirect prompt injection, the risk is mitigated by the mandatory human review step and the use of structured templates for persona drafting.
[COMMAND_EXECUTION]: No dangerous system commands, privilege escalation attempts, or unauthorized shell executions were detected. The file-writing operations are scoped to a specific documentation directory.

recipe-persona