The Agent Skills Directory

[SAFE]: The skill follows least-privilege principles by using native agent tools (like image_gen) for generative image editing without requesting unnecessary file system or network permissions.
[SAFE]: Data privacy is explicitly addressed in the workflow, instructing the agent never to publish or upload user portraits to public repositories without separate user consent.
[SAFE]: The skill includes robust identity-preservation instructions and a quality gate mechanism to prevent the generation of synthetic or deceptive imagery that deviates from the original subject.
[SAFE]: The instructional logic for handling user feedback includes a safety constraint that prevents automatic external writing or submission, ensuring human-in-the-loop validation for any permanent rule changes.

lov-professional-portrait