hive-mind-advanced

foryourhealth111-pixel/Vibe-SkillsCoding AgentsReview advisedSnapshot 6fac07fce8f7

Installs: 0

Summary

Advanced Hive Mind collective intelligence system for queen-led multi-agent coordination with consensus mechanisms and persistent memory

SKILL.md

Hive Mind Advanced Skill

Master the advanced Hive Mind collective intelligence system for sophisticated multi-agent coordination using queen-led architecture, Byzantine consensus, and collective memory.

Overview

The Hive Mind system represents the pinnacle of multi-agent coordination in Claude Flow, implementing a queen-led hierarchical architecture where a strategic queen coordinator directs specialized worker agents through collective decision-making and shared memory.

Core Concepts

Architecture Patterns

Queen-Led Coordination

Strategic queen agents orchestrate high-level objectives
Tactical queens manage mid-level execution
Adaptive queens dynamically adjust strategies based on performance

Worker Specialization

Researcher agents: Analysis and investigation
Coder agents: Implementation and development
Analyst agents: Data processing and metrics
Tester agents: Quality assurance and validation
Architect agents: System design and planning
Reviewer agents: Code review and improvement
Optimizer agents: Performance enhancement
Documenter agents: Documentation generation

Collective Memory System

Shared knowledge base across all agents
LRU cache with memory pressure handling
SQLite persistence with WAL mode
Memory consolidation and association
Access pattern tracking and optimization

Consensus Mechanisms

Majority Consensus Simple voting where the option with most votes wins.

Weighted Consensus Queen vote counts as 3x weight, providing strategic guidance.

Byzantine Fault Tolerance Requires 2/3 majority for decision approval, ensuring robust consensus even with faulty agents.

# Basic initialization
npx claude-flow hive-mind init

# Force reinitialize
npx claude-flow hive-mind init --force

# Custom configuration
npx claude-flow hive-mind init --config hive-config.json

# Basic spawn with objective
npx claude-flow hive-mind spawn "Build microservices architecture"

# Strategic queen type
npx claude-flow hive-mind spawn "Research AI patterns" --queen-type strategic

# Tactical queen with max workers
npx claude-flow hive-mind spawn "Implement API" --queen-type tactical --max-workers 12

# Adaptive queen with consensus
npx claude-flow hive-mind spawn "Optimize system" --queen-type adaptive --consensus byzantine

# Generate Claude Code commands
npx claude-flow hive-mind spawn "Build full-stack app" --claude

# Check hive mind status
npx claude-flow hive-mind status

# Get detailed metrics
npx claude-flow hive-mind metrics

# Monitor collective memory
npx claude-flow hive-mind memory

# List active sessions
npx claude-flow hive-mind sessions

# Pause a session
npx claude-flow hive-mind pause <session-id>

# Resume a paused session
npx claude-flow hive-mind resume <session-id>

# Stop a running session
npx claude-flow hive-mind stop <session-id>

// Programmatic consensus building
const decision = await hiveMind.buildConsensus(
  'Architecture pattern selection',
  ['microservices', 'monolith', 'serverless']
);

// Result includes:
// - decision: Winning option
// - confidence: Vote percentage
// - votes: Individual agent votes

// Store in collective memory
await memory.store('api-patterns', {
  rest: { pros: [...], cons: [...] },
  graphql: { pros: [...], cons: [...] }
}, 'knowledge', { confidence: 0.95 });

// Search memory by pattern
const results = await memory.search('api*', {
  type: 'knowledge',
  minConfidence: 0.8,
  limit: 50
});

// Get related memories
const related = await memory.getRelated('api-patterns', 10);

// Build associations
await memory.associate('rest-api', 'authentication', 0.9);

// Create task (auto-assigned)
const task = await hiveMind.createTask(
  'Implement user authentication',
  priority: 8,
  { estimatedDuration: 30000 }
);

// Configure auto-scaling
const config = {
  autoScale: true,
  maxWorkers: 12,
  scaleUpThreshold: 2, // Pending tasks per idle worker
  scaleDownThreshold: 2 // Idle workers above pending tasks
};

npx claude-flow hive-mind spawn "Build REST API" --claude

Task("Queen Coordinator", "Orchestrate REST API development...", "coordinator")
Task("Backend Developer", "Implement Express routes...", "backend-dev")
Task("Database Architect", "Design PostgreSQL schema...", "code-analyzer")
Task("Test Engineer", "Create Jest test suite...", "tester")

# Use hive mind for SPARC workflow
npx claude-flow sparc tdd "User authentication" --hive-mind

# Spawns:
# - Specification agent
# - Architecture agent
# - Coder agents
# - Tester agents
# - Reviewer agents

# Repository analysis with hive mind
npx claude-flow hive-mind spawn "Analyze repo quality" --objective "owner/repo"

# PR review coordination
npx claude-flow hive-mind spawn "Review PR #123" --queen-type tactical

// Get performance insights
const insights = hiveMind.getPerformanceInsights();

// Includes:
// - asyncQueue utilization
// - Batch processing stats
// - Success rates
// - Average processing times
// - Memory efficiency

{
  "objective": "Build microservices",
  "name": "my-hive",
  "queenType": "strategic", // strategic | tactical | adaptive
  "maxWorkers": 8,
  "consensusAlgorithm": "byzantine", // majority | weighted | byzantine
  "autoScale": true,
  "memorySize": 100, // MB
  "taskTimeout": 60, // minutes
  "encryption": false
}

{
  "maxSize": 100, // MB
  "compressionThreshold": 1024, // bytes
  "gcInterval": 300000, // 5 minutes
  "cacheSize": 1000,
  "cacheMemoryMB": 50,
  "enablePooling": true,
  "enableAsyncOperations": true
}

npx claude-flow hive-mind spawn "Research ML frameworks" --queen-type strategic

npx claude-flow hive-mind spawn "Build authentication" --queen-type tactical

npx claude-flow hive-mind spawn "Optimize performance" --queen-type adaptive

// After successful pattern implementation
await memory.store('auth-pattern', {
  approach: 'JWT with refresh tokens',
  pros: ['Stateless', 'Scalable'],
  cons: ['Token size', 'Revocation complexity'],
  implementation: {...}
}, 'knowledge', { confidence: 0.95 });

// Link related concepts
await memory.associate('jwt-auth', 'refresh-tokens', 0.9);
await memory.associate('jwt-auth', 'oauth2', 0.7);

# Regular status checks
npx claude-flow hive-mind status

# Track metrics
npx claude-flow hive-mind metrics

# Analyze memory usage
npx claude-flow hive-mind memory

// Create checkpoints at key milestones
await sessionManager.saveCheckpoint(
  sessionId,
  'api-routes-complete',
  { completedRoutes: [...], remaining: [...] }
);

# Resume from any previous state
npx claude-flow hive-mind resume <session-id>

# Run garbage collection
npx claude-flow hive-mind memory --gc

# Optimize database
npx claude-flow hive-mind memory --optimize

# Export and clear
npx claude-flow hive-mind memory --export --clear

// Increase cache size in config
{
  "cacheSize": 2000,
  "cacheMemoryMB": 100
}

// Enable worker type caching
// The system caches best worker matches for 5 minutes
// Automatic - no configuration needed

// Increase async queue concurrency
{
  "asyncQueueConcurrency": 20 // Default: min(maxWorkers * 2, 20)
}

# Switch to weighted consensus for more decisive results
npx claude-flow hive-mind spawn "..." --consensus weighted

# Or use simple majority
npx claude-flow hive-mind spawn "..." --consensus majority

name: security-auditor
type: specialist
capabilities:
  - vulnerability-scanning
  - security-review
  - penetration-testing
  - compliance-checking
priority: high

// Automatic pattern learning
// Happens after successful task completion
// Stores in collective memory
// Improves future task matching

# Frontend hive
npx claude-flow hive-mind spawn "Build UI" --name frontend-hive

# Backend hive
npx claude-flow hive-mind spawn "Build API" --name backend-hive

# They share collective memory for coordination

# Export session for backup
npx claude-flow hive-mind export <session-id> --output backup.json

# Import session
npx claude-flow hive-mind import backup.json

const hiveMind = new HiveMindCore({
  objective: 'Build system',
  queenType: 'strategic',
  maxWorkers: 8,
  consensusAlgorithm: 'byzantine'
});

await hiveMind.initialize();
await hiveMind.spawnQueen(queenData);
await hiveMind.spawnWorkers(['coder', 'tester']);
await hiveMind.createTask('Implement feature', 7);
const decision = await hiveMind.buildConsensus('topic', options);
const status = hiveMind.getStatus();
await hiveMind.shutdown();

const memory = new CollectiveMemory({
  swarmId: 'hive-123',
  maxSize: 100,
  cacheSize: 1000
});

await memory.store(key, value, type, metadata);
const data = await memory.retrieve(key);
const results = await memory.search(pattern, options);
const related = await memory.getRelated(key, limit);
await memory.associate(key1, key2, strength);
const stats = memory.getStatistics();
const analytics = memory.getAnalytics();
const health = await memory.healthCheck();

const sessionManager = new HiveMindSessionManager();

const sessionId = await sessionManager.createSession(
  swarmId, swarmName, objective, metadata
);

await sessionManager.saveCheckpoint(sessionId, name, data);
const sessions = await sessionManager.getActiveSessions();
const session = await sessionManager.getSession(sessionId);
await sessionManager.pauseSession(sessionId);
await sessionManager.resumeSession(sessionId);
await sessionManager.stopSession(sessionId);
await sessionManager.completeSession(sessionId);

# Initialize hive mind
npx claude-flow hive-mind init

# Spawn full-stack hive
npx claude-flow hive-mind spawn "Build e-commerce platform" \
  --queen-type strategic \
  --max-workers 10 \
  --consensus weighted \
  --claude

# Output generates Claude Code commands:
# - Queen coordinator
# - Frontend developers (React)
# - Backend developers (Node.js)
# - Database architects
# - DevOps engineers
# - Security auditors
# - Test engineers
# - Documentation specialists

# Spawn research hive
npx claude-flow hive-mind spawn "Research GraphQL vs REST" \
  --queen-type adaptive \
  --consensus byzantine

# Researchers gather data
# Analysts process findings
# Queen builds consensus on recommendation
# Results stored in collective memory

# Review coordination
npx claude-flow hive-mind spawn "Review PR #456" \
  --queen-type tactical \
  --max-workers 6

# Spawns:
# - Code analyzers
# - Security reviewers
# - Performance reviewers
# - Test coverage analyzers
# - Documentation reviewers
# - Consensus on approval/changes

hive-mind-advanced

Summary

Hive Mind Advanced Skill

Overview

Core Concepts

Architecture Patterns

Consensus Mechanisms

Getting Started

1. Initialize Hive Mind

2. Spawn a Swarm

3. Monitor Status

Advanced Workflows

Session Management

Consensus Building

Collective Memory

Task Distribution

Integration Patterns

With Claude Code

With SPARC Methodology

With GitHub Integration

Performance Optimization

Memory Optimization

Performance Metrics

Task Execution

Configuration

Hive Mind Config

Memory Config

Hooks Integration

Best Practices

1. Choose the Right Queen Type

2. Leverage Consensus

3. Utilize Collective Memory

4. Monitor Performance

5. Session Management

Troubleshooting

Memory Issues

Performance Issues

Consensus Failures

Advanced Topics

Custom Worker Types

Neural Pattern Training

Multi-Hive Coordination

Export/Import Sessions

API Reference

HiveMindCore

CollectiveMemory

HiveMindSessionManager

Examples

Full-Stack Development

Research and Analysis

Code Review

Skill Progression

Beginner

Intermediate

Advanced

Related Skills

References

Related skills