feature-discussion

Multi-agent feature discussion coordinates persona-driven agents through five steps—purpose, alternatives, scope, requirements, and UI prompts—to debate, validate scope, and surface risks. It outputs session.json, discussion_log.md, and step deliverables with Devil's Advocate rounds and quality gates.

A devil's advocate round in feature ideation is a structured debate phase where agents present opposing views and challenge assumptions. It ensures cross-functional consensus by forcing agents to dispute proposals before passing quality gates and moving to the next step.

UX analysis and behavioral psychology input are structured by assigning dedicated agent personas to evaluate user needs and behavioral triggers. They debate with other agents during each step to align requirements, generating UI prompts and step deliverables for implementation.

Yes, multi-agent feature discussion integrates external domain docs and codebase research to ground decisions. Agents reference these sources during the five-step workflow to validate purpose, alternatives, scope, requirements, and UI prompts against real project constraints.

The best way to align cross-functional teams on feature scope is a stepwise multi-agent debate where personas evaluate purpose, alternatives, and requirements. Real-time dispute and synthesis with user-involved decisions ensure consensus and produce traceable artifacts.

You need a structured feature discussion with agent teams when cross-functional consensus and risk surfacing are critical for new features. It is necessary when standard planning lacks persona-driven evaluation, real-time dispute, and traceable artifact generation for implementation.