OpenClaw Consensus

Wraps the OpenClaw Consensus API to run the same prompt across up to nine language models and return aggregated answers with confidence scores. You get three tools: consensus() for multi-model agreement, disagreement_score() to surface when models diverge, and cheapest_route() to try fast/balanced/deep tiers until confidence hits your threshold. Useful when you want a second opinion before accepting a risky code review, routing decision, or factual claim. Requires a RapidAPI key for the OpenClaw endpoint. The consensus output includes vote breakdowns and model metadata so you can see which responses aligned and where outliers appeared.