hive-mind-advanced

The Hive Mind Advanced skill is a sophisticated orchestration layer for Claude Code, designed to manage high-complexity software engineering workflows through a hierarchical, queen-led agent structure. By delegating tasks to specialized agents—including researchers, coders, architects, and testers—the system ensures high-quality execution and strategic alignment. It is ideal for software teams and developers managing large-scale, multi-stage projects that require automated coordination, fault-tolerant decision-making, and long-term knowledge retention. The architecture leverages persistent SQLite-backed collective memory to maintain context across sessions, enabling agents to learn from historical patterns and association data.

• Advanced Queen-Led Coordination: Hierarchical management using strategic, tactical, and adaptive queen agents to direct worker agents.

• Robust Consensus Mechanisms: Implements Majority, Weighted, and Byzantine Fault Tolerance (BFT) algorithms to validate decisions and minimize agent errors.

• Persistent Collective Memory: Includes an LRU-cached, SQLite-persisted knowledge base for task-specific data, errors, metrics, and long-term insights.

• Intelligent Auto-Scaling: Dynamically adjusts worker counts based on pending tasks and system load, ensuring optimal resource allocation.

• Specialization Framework: Integrates with 74+ specialized agent roles ranging from documenters and reviewers to security auditors and test engineers.

• Session Lifecycle Management: Full support for pausing, resuming, and checkpointing complex workflows with export/import capabilities.

• Utilize the hive-mind spawn command to initiate swarms for specific objectives, such as building microservices or optimizing legacy codebases.

• Configure memory types (knowledge, task, consensus, system) to ensure relevant data persistence and retrieval using search, getRelated, and associate methods.

• Monitor operational performance via the metrics and status commands, providing real-time visibility into swarm health and consensus confidence scores.

• Input requirements include high-level project objectives, while outputs consist of structured execution plans, code artifacts, test reports, and optimized architectural decisions.

• Note that this skill requires the Claude Flow or Ruflo environment to manage the underlying plugin hooks and WASM-powered policy engines effectively.