hive-mind-advanced

Multi-agent coordination uses a queen-led architecture where a strategic coordinator directs specialized worker agents (researchers, coders, testers, architects) to collaborate on long-running projects. This Skill implements hierarchical queen coordination with Byzantine consensus to ensure robust decision-making across the swarm, enabling coordinated execution even when some agents fail or disagree.

Byzantine consensus is a fault-tolerant decision-making protocol that reaches agreement even when some agents are faulty or behave unpredictably, requiring a 2/3 majority to approve critical decisions. It protects swarm coordination from individual agent failures, ensuring robust outcomes in systems where perfect agent reliability cannot be guaranteed.

Collective memory systems use LRU caching layered with SQLite WAL persistence to store and retrieve shared knowledge persistently. This Skill provides memory types with configurable TTLs, search and association capabilities, and programmatic APIs so all agents access consistent, up-to-date context across distributed sessions without manual synchronization.

Yes. Spawn a queen-led swarm with specialized worker roles—architects for design, coders for implementation, testers for validation—to collaboratively develop and validate microservices systems. The swarm leverages Byzantine consensus for architectural decisions and collective memory to maintain shared design knowledge and integration points.

This Skill includes auto-scaling controls that adjust the swarm size based on workload demands, allowing agents to spawn or retire as needed. Memory-based task assignment routes work to available agents dynamically, and the queen coordinator adapts consensus thresholds and worker allocation automatically.

Yes. It's designed for multi-role projects where specialized agents (researcher, coder, analyst, tester, architect, reviewer, optimizer, documenter) work together over extended periods. Shared memory with TTL-managed memory types, adaptive consensus, and dynamic task distribution enable agents to maintain state and coordinate complex workflows across multiple sessions.