Consensus MCP Board – Decision firewall for AI code reviews and PR merges

Summary: Consensus MCP Board is an open source MCP server and workflow UI that enforces consensus checks and decision firewalls by requiring multiple AI models or agents to agree before executing actions. It enhances reliability in AI-driven code reviews and PR merges by combining weighted consensus, reputation scores, risk thresholds, and optional human approval.

What it does

It runs multiple specialized AI agents in parallel to evaluate code changes and resolves outcomes through weighted consensus, preventing single-agent errors. The system operates fully locally using SQLite and localhost without cloud dependencies.

Who it's for

Developers and teams using AI for code reviews, PR workflows, or agent systems needing improved decision reliability and risk mitigation.

Why it matters

It reduces risks from inconsistent or hallucinated AI decisions by requiring agreement among multiple models before merging code or triggering actions.