skill-debate-integration

Multi-model AI debates for design reviews are automated by integrating session-aware storage, quality gates, and export workflows into Claude-Octopus. Structured adversarial debates among Gemini, Codex, and Claude are initiated via the /debate command.

Quality gates in multi-model AI debate workflows enforce per-round scoring on each advisor response to ensure outcomes are substantive, sourced, and engaged. This guarantees auditable and cost-conscious discussions across all debate rounds.

Token cost analytics across AI debate rounds are tracked automatically by the integration's built-in cost monitoring. It logs token usage and debate expenses per round, providing visibility into multi-model discussion costs within production workflows.

AI debate transcripts and synthesis results can be exported to PPTX, DOCX, PDF, or Markdown formats. The document export integration generates these deliverables directly from the session-aware storage after the debate concludes.

Session-aware storage organizes AI debate transcripts by Claude Code session and topic slug, creating structured directories. This ensures multi-model discussions remain auditable and easily retrievable within production workflows.

Claude-Octopus is required as the base environment because the debate integration directly extends its workflows. It applies session-based directories, quality gates, and export interfaces specifically within the Claude-Octopus ecosystem.