Cross-model adversarial review and idea generation for Claude Code. A Claude session orchestrates and ratifies; OpenAI (codex), Moonshot (kimi) and xAI (grok) seats probe, replicate and generate — all on subscription auth, no API key. Ten modes: review, forge, instrument audit, fragment verify, blind replication, and more.
Command
Part of concilium
Run a Concilium BRAINSTORMING campaign — open-book cross-model idea brainstorminging with a live read-only copy of the project's database per seat. Findings replayed, ideas flow to human-reviewed dossiers, never orchestrator-curated.
Command
Part of concilium
Run a Concilium FORGE round — cross-model idea generation against an open question, with a shared register. Nothing is judged or voted on.
Command
Part of concilium
Route a claim, or the uncommitted diff, to the cross-model Concilium reviewer.