design-sub-agents

Design adversarial reviewer sub-agents by defining distinct personas that critique outputs from creation skills, ensuring edge cases and biases are challenged. Wire these personas into the trigger map to validate agent outputs.

An adversarial reviewer persona is a distinct lens in prompt design that protects outputs by challenging assumptions and catching subtle errors. It introduces adversarial perspectives to quality gates, ensuring outputs are validated against edge cases.

Introduce sub-agents after defining audiences and designing skills, when domain-input/audiences.md has sub-agent candidates. These reviewer personas validate outputs by mapping to sub-agent design templates and defining trigger integration.

Map reviewer sub-agents by defining trigger integration and evaluation priorities that align with the sub-agents directory for the target OS. This structured workflow ensures personas map to audiences and validate creation skill outputs.

Draft 2-4 reviewer personas with distinct evaluation lenses to establish effective quality gates. Wire these adversarial reviewer sub-agents into the trigger map after designing skills to challenge output assumptions and catch biases.

Yes, adversarial sub-agents validate outputs from creation skills by introducing reviewer lenses that challenge assumptions and catch subtle errors. They act as quality gates, ensuring outputs are tested for edge cases and biases.