Babel · dt-tb-labs bundle Use when the user invokes /babel, or asks to orchestrate multiple models for a dev task ("多モデルで開発", "develop with multiple models", "5モデルで", "with 5 models", "複数AIでレビューさせて", "have multiple AIs review it"), or asks for a measured improvement loop ("ループエンジニアリング", "loop engineering", "スコアを上げるまで回して", "iterate until the benchmark improves", "optimize against a metric"). Orchestrates 5 models (Fable5/Opus/Sonnet/GPT-5.6-SOL/agy) across the superpowers pipeline (brainstorm→plan→implement→review), injecting multi-model debate/build-debug/acceptance-gate into each phase to exceed single-frontier-model quality (multi-model ensemble pattern). Routes measurable search tasks into a multi-model optimization loop against an automated oracle, and proposes a Workflow fan-out when the task is many independent sites. Not for JavaScript Babel transpiler tasks.
