What (One-liner)

Canvas multi-agent institution debate: N LLMs with different identities debate 3 rounds → consensus extraction → CIDE-6.1 InstitutionDecision contract.

How to use

Institution template (fedFOMC, 12 members) → institution_committee
                                            → {stance: hawkish, consensusReached: true, assetViews: [SPY-, TLT+]}

Core formulas

See frontmatter. 3-round LLM debate → rule-based consensus extraction → InstitutionDecision.

Assumptions & applicability

Assumptions: ≥2 agents + Phase K complete + sufficient LLM quota.

Applies to: FOMC/MPC/IC simulation, multi-party decision demo, BoE herding rebuttal evidence.

Does not apply to: single agent, quantitative consensus, byte-deterministic, live.

Known limitations

See frontmatter. Six items: LLM cost / fragile consensus rule / herding risk / subjective biases / WebSocket fallback / keyFactors hallucination.

References

• Li et al. (2023) — Multi-agent debate
• BoE (2025-04) — AI herding stress test
• Pangura ExplainableAIService + CIDE-6.1 contract

Golden Test

Tier 5 attestation — pending Phase B framework.

Changelog

• 1.0.0 (2026-04-21) — First Active (Tier 5 track)