Digital Twin

This skill appears coherent with its purpose, but review these before installing or using: - Consent & legality: The skill can produce an installable persona that mimics a real person. Only build personas for people who have given explicit consent; check company policy and local law about impersonation and personal data processing. - Trust the connector: The skill depends on your separate Fireflies connector to fetch transcripts. Ensure that connector is properly scoped, audited, and only returns the transcripts you expect. The digital-twin skill's claim that it won't persist or transmit raw transcripts is an instruction — verify behavior by testing with your own transcripts first. - Secrets & access: This skill asks for no creds itself, but the Fireflies connector you use will have credentials — follow least privilege for that connector and review its logs/permissions. - Review generated output: The skill will produce an installable personality SKILL.md and supporting files that enable impersonation. Before installing a generated persona into production or enabling it persistently, inspect the generated skill for overly broad behaviors, sensitive content inclusion, or scope creep. - Start small: Test the workflow using transcripts of yourself or consenting colleagues, inspect outputs, and confirm you can revoke or remove any generated persona. If you want a stronger assessment, provide the untruncated parts of SKILL.md (the runtime workflow was truncated) or example transcripts and a generated persona so I can check whether any steps read or transmit unexpected paths, environment variables, or external endpoints.