hive-mind-advanced

Multi-agent coordination uses hierarchical orchestration where a leader agent distributes specialized tasks to worker agents, collects results, and reaches decisions through consensus. This skill automates that workflow for tasks like research, coding, testing, and architecture—enabling parallel execution and shared memory across agents.

Byzantine consensus is a decision-making algorithm that reaches agreement even when some agents fail or provide incorrect information. For agent swarms, it ensures robust architectural and technical decisions survive individual agent errors, critical when coordinating fault-tolerant systems.

Yes. This skill provides a collective memory system backed by SQLite with WAL mode, LRU cache, and TTLs. All agents in the swarm access the same knowledge base, eliminating redundant work and enabling learnings from one task to inform subsequent decisions.

Initialize the Hive Mind system with `npx claude-flow hive-mind init`, then spawn a swarm with `npx claude-flow hive-mind spawn "[your task]"`. The system automatically assigns specialized roles—researcher, coder, tester, architect—distributes work in parallel, and manages consensus decisions.

The skill applies configurable consensus algorithms: majority voting, weighted voting, or Byzantine consensus. Failed or dissenting agents are quarantined; the system logs disagreements and proceeds with the consensus result, ensuring forward progress in multi-agent workflows.

Yes. While designed for development (research, coding, testing, architecture, documentation, optimization), the consensus and task-distribution framework generalizes to any domain requiring multi-agent coordination and collective decision-making.