CHP: Open-source Consensus Hardening Protocol for preventing sycophantic convergence in multi-agent LLM systems (opens in new tab)

Consensus Hardening Protocol — decision-governance layer for multi-agent AI: foundation disclosure, adversarial attack, R0 gate, and EXPLORING → PROVISIONAL_LOCK → LOCKED progression with auditable cross-model payload envelopes.