When NOT to Use This Skill

• Local development without cloud infrastructure needs
• Simple scripts that do not require sandboxed execution
• Operations without distributed computing requirements
• Tasks that can run on single-machine environments

Success Criteria

• API response time: <200ms for sandbox creation
• Deployment success rate: >99%
• Sandbox startup time: <5s
• Network latency: <50ms between sandboxes
• Resource utilization: <80% CPU/memory per sandbox
• Uptime: >99.9% for production deployments

Edge Cases & Error Handling

• Rate Limits: Flow Nexus API has request limits; implement queuing and backoff
• Authentication Failures: Validate API tokens before operations; refresh expired tokens
• Network Issues: Retry failed requests with exponential backoff (max 5 retries)
• Quota Exhaustion: Monitor sandbox/compute quotas; alert before limits
• Sandbox Timeouts: Set appropriate timeout values; clean up orphaned sandboxes
• Deployment Failures: Implement rollback strategies; maintain previous working state

Guardrails & Safety

• NEVER expose API keys or authentication tokens in code or logs
• ALWAYS validate responses from Flow Nexus API before processing
• ALWAYS implement timeout limits for long-running operations
• NEVER trust user input for sandbox commands without validation
• ALWAYS monitor resource usage to prevent runaway processes
• ALWAYS clean up sandboxes and resources after task completion

Evidence-Based Validation

• Verify platform health: Check Flow Nexus status endpoint before operations
• Validate deployments: Test sandbox connectivity and functionality
• Monitor costs: Track compute usage and spending against budgets
• Test failure scenarios: Simulate network failures, timeouts, auth errors
• Benchmark performance: Compare actual vs expected latency/throughput

Flow Nexus Neural Networks

Deploy, train, and manage neural networks in distributed E2B sandbox environments. Train custom models with multiple architectures (feedforward, LSTM, GAN, transformer) or use pre-built templates from the marketplace.

Prerequisites

# Add Flow Nexus MCP server
claude mcp add flow-nexus npx flow-nexus@latest mcp start

# Register and login
npx flow-nexus@latest register
npx flow-nexus@latest login

Core Capabilities

1. Single-Node Neural Training

Train neural networks with custom architectures and configurations.

Available Architectures:

• feedforward - Standard fully-connected networks
• lstm - Long Short-Term Memory for sequences
• gan - Generative Adversarial Networks
• autoencoder - Dimensionality reduction
• transformer - Attention-based models

Training Tiers:

• nano - Minimal resources (fast, limited)
• mini - Small models
• small - Standard models
• medium - Complex models
• large - Large-scale training

Example: Train Custom Classifier

mcp__flow-nexus__neural_train({
  config: {
    architecture: {
      type: "feedforward",
      layers: [
        { type: "dense", units: 256, activation: "relu" },
        { type: "dropout", rate: 0.3 },
        { type: "dense", units: 128, activation: "relu" },
        { type: "dropout", rate: 0.2 },
        { type: "dense", units: 64, activation: "relu" },
        { type: "dense", units: 10, activation: "softmax" }
      ]
    },
    training: {
      epochs: 100,
      batch_size: 32,
      learning_rate: 0.001,
      optimizer: "adam"
    },
    divergent: {
      enabled: true,
      pattern: "lateral", // quantum, chaotic, associative, evolutionary
      factor: 0.5
    }
  },
  tier: "small",
  user_id: "your_user_id"
})

Example: LSTM for Time Series

mcp__flow-nexus__neural_train({
  config: {
    architecture: {
      type: "lstm",
      layers: [
        { type: "lstm", units: 128, return_sequences: true },
        { type: "dropout", rate: 0.2 },
        { type: "lstm", units: 64 },
        { type: "dense", units: 1, activation: "linear" }
      ]
    },
    training: {
      epochs: 150,
      batch_size: 64,
      learning_rate: 0.01,
      optimizer: "adam"
    }
  },
  tier: "medium"
})

Example: Transformer Architecture

mcp__flow-nexus__neural_train({
  config: {
    architecture: {
      type: "transformer",
      layers: [
        { type: "embedding", vocab_size: 10000, embedding_dim: 512 },
        { type: "transformer_encoder", num_heads: 8, ff_dim: 2048 },
        { type: "global_average_pooling" },
        { type: "dense", units: 128, activation: "relu" },
        { type: "dense", units: 2, activation: "softmax" }
      ]
    },
    training: {
      epochs: 50,
      batch_size: 16,
      learning_rate: 0.0001,
      optimizer: "adam"
    }
  },
  tier: "large"
})

2. Model Inference

Run predictions on trained models.

mcp__flow-nexus__neural_predict({
  model_id: "model_abc123",
  input: [
    [0.5, 0.3, 0.2, 0.1],
    [0.8, 0.1, 0.05, 0.05],
    [0.2, 0.6, 0.15, 0.05]
  ],
  user_id: "your_user_id"
})

Response:

{
  "predictions": [
    [0.12, 0.85, 0.03],
    [0.89, 0.08, 0.03],
    [0.05, 0.92, 0.03]
  ],
  "inference_time_ms": 45,
  "model_version": "1.0.0"
}

3. Template Marketplace

Browse and deploy pre-trained models from the marketplace.

List Available Templates

mcp__flow-nexus__neural_list_templates({
  category: "classification", // timeseries, regression, nlp, vision, anomaly, generative
  tier: "free", // or "paid"
  search: "sentiment",
  limit: 20
})

Response:

{
  "templates": [
    {
      "id": "sentiment-analysis-v2",
      "name": "Sentiment Analysis Classifier",
      "description": "Pre-trained BERT model for sentiment analysis",
      "category": "nlp",
      "accuracy": 0.94,
      "downloads": 1523,
      "tier": "free"
    },
    {
      "id": "image-classifier-resnet",
      "name": "ResNet Image Classifier",
      "description": "ResNet-50 for image classification",
      "category": "vision",
      "accuracy": 0.96,
      "downloads": 2341,
      "tier": "paid"
    }
  ]
}

Deploy Template

mcp__flow-nexus__neural_deploy_template({
  template_id: "sentiment-analysis-v2",
  custom_config: {
    training: {
      epochs: 50,
      learning_rate: 0.0001
    }
  },
  user_id: "your_user_id"
})

4. Distributed Training Clusters

Train large models across multiple E2B sandboxes with distributed computing.

Initialize Cluster

mcp__flow-nexus__neural_cluster_init({
  name: "large-model-cluster",
  architecture: "transformer", // transformer, cnn, rnn, gnn, hybrid
  topology: "mesh", // mesh, ring, star, hierarchical
  consensus: "proof-of-learning", // byzantine, raft, gossip
  daaEnabled: true, // Decentralized Autonomous Agents
  wasmOptimization: true
})

Response:

{
  "cluster_id": "cluster_xyz789",
  "name": "large-model-cluster",
  "status": "initializing",
  "topology": "mesh",
  "max_nodes": 100,
  "created_at": "2025-10-19T10:30:00Z"
}

Deploy Worker Nodes

// Deploy parameter server
mcp__flow-nexus__neural_node_deploy({
  cluster_id: "cluster_xyz789",
  node_type: "parameter_server",
  model: "large",
  template: "nodejs",
  capabilities: ["parameter_management", "gradient_aggregation"],
  autonomy: 0.8
})

// Deploy worker nodes
mcp__flow-nexus__neural_node_deploy({
  cluster_id: "cluster_xyz789",
  node_type: "worker",
  model: "xl",
  role: "worker",
  capabilities: ["training", "inference"],
  layers: [
    { type: "transformer_encoder", num_heads: 16 },
    { type: "feed_forward", units: 4096 }
  ],
  autonomy: 0.9
})

// Deploy aggregator
mcp__flow-nexus__neural_node_deploy({
  cluster_id: "cluster_xyz789",
  node_type: "aggregator",
  model: "large",
  capabilities: ["gradient_aggregation", "model_synchronization"]
})

Connect Cluster Topology

mcp__flow-nexus__neural_cluster_connect({
  cluster_id: "cluster_xyz789",
  topology: "mesh" // Override default if needed
})

Start Distributed Training

mcp__flow-nexus__neural_train_distributed({
  cluster_id: "cluster_xyz789",
  dataset: "imagenet", // or custom dataset identifier
  epochs: 100,
  batch_size: 128,
  learning_rate: 0.001,
  optimizer: "adam", // sgd, rmsprop, adagrad
  federated: true // Enable federated learning
})

Federated Learning Example:

mcp__flow-nexus__neural_train_distributed({
  cluster_id: "cluster_xyz789",
  dataset: "medical_images_distributed",
  epochs: 200,
  batch_size: 64,
  learning_rate: 0.0001,
  optimizer: "adam",
  federated: true, // Data stays on local nodes
  aggregation_rounds: 50,
  min_nodes_per_round: 5
})

Monitor Cluster Status

mcp__flow-nexus__neural_cluster_status({
  cluster_id: "cluster_xyz789"
})

Response:

{
  "cluster_id": "cluster_xyz789",
  "status": "training",
  "nodes": [
    {
      "node_id": "node_001",
      "type": "parameter_server",
      "status": "active",
      "cpu_usage": 0.75,
      "memory_usage": 0.82
    },
    {
      "node_id": "node_002",
      "type": "worker",
      "status": "active",
      "training_progress": 0.45
    }
  ],
  "training_metrics": {
    "current_epoch": 45,
    "total_epochs": 100,
    "loss": 0.234,
    "accuracy": 0.891
  }
}

Run Distributed Inference

mcp__flow-nexus__neural_predict_distributed({
  cluster_id: "cluster_xyz789",
  input_data: JSON.stringify([
    [0.1, 0.2, 0.3],
    [0.4, 0.5, 0.6]
  ]),
  aggregation: "ensemble" // mean, majority, weighted, ensemble
})

Terminate Cluster

mcp__flow-nexus__neural_cluster_terminate({
  cluster_id: "cluster_xyz789"
})

5. Model Management

List Your Models

mcp__flow-nexus__neural_list_models({
  user_id: "your_user_id",
  include_public: true
})

Response:

{
  "models": [
    {
      "model_id": "model_abc123",
      "name": "Custom Classifier v1",
      "architecture": "feedforward",
      "accuracy": 0.92,
      "created_at": "2025-10-15T14:20:00Z",
      "status": "trained"
    },
    {
      "model_id": "model_def456",
      "name": "LSTM Forecaster",
      "architecture": "lstm",
      "mse": 0.0045,
      "created_at": "2025-10-18T09:15:00Z",
      "status": "training"
    }
  ]
}

Check Training Status

mcp__flow-nexus__neural_training_status({
  job_id: "job_training_xyz"
})

Response:

{
  "job_id": "job_training_xyz",
  "status": "training",
  "progress": 0.67,
  "current_epoch": 67,
  "total_epochs": 100,
  "current_loss": 0.234,
  "estimated_completion": "2025-10-19T12:45:00Z"
}

Performance Benchmarking

mcp__flow-nexus__neural_performance_benchmark({
  model_id: "model_abc123",
  benchmark_type: "comprehensive" // inference, throughput, memory, comprehensive
})

Response:

{
  "model_id": "model_abc123",
  "benchmarks": {
    "inference_latency_ms": 12.5,
    "throughput_qps": 8000,
    "memory_usage_mb": 245,
    "gpu_utilization": 0.78,
    "accuracy": 0.92,
    "f1_score": 0.89
  },
  "timestamp": "2025-10-19T11:00:00Z"
}

Create Validation Workflow

mcp__flow-nexus__neural_validation_workflow({
  model_id: "model_abc123",
  user_id: "your_user_id",
  validation_type: "comprehensive" // performance, accuracy, robustness, comprehensive
})

6. Publishing and Marketplace

Publish Model as Template

mcp__flow-nexus__neural_publish_template({
  model_id: "model_abc123",
  name: "High-Accuracy Sentiment Classifier",
  description: "Fine-tuned BERT model for sentiment analysis with 94% accuracy",
  category: "nlp",
  price: 0, // 0 for free, or credits amount
  user_id: "your_user_id"
})

Rate a Template

mcp__flow-nexus__neural_rate_template({
  template_id: "sentiment-analysis-v2",
  rating: 5,
  review: "Excellent model! Achieved 95% accuracy on my dataset.",
  user_id: "your_user_id"
})

Common Use Cases

Image Classification with CNN

// Initialize cluster for large-scale image training
const cluster = await mcp__flow-nexus__neural_cluster_init({
  name: "image-classification-cluster",
  architecture: "cnn",
  topology: "hierarchical",
  wasmOptimization: true
})

// Deploy worker nodes
await mcp__flow-nexus__neural_node_deploy({
  cluster_id: cluster.cluster_id,
  node_type: "worker",
  model: "large",
  capabilities: ["training", "data_augmentation"]
})

// Start training
await mcp__flow-nexus__neural_train_distributed({
  cluster_id: cluster.cluster_id,
  dataset: "custom_images",
  epochs: 100,
  batch_size: 64,
  learning_rate: 0.001,
  optimizer: "adam"
})

NLP Sentiment Analysis

// Use pre-built template
const deployment = await mcp__flow-nexus__neural_deploy_template({
  template_id: "sentiment-analysis-v2",
  custom_config: {
    training: {
      epochs: 30,
      batch_size: 16
    }
  }
})

// Run inference
const result = await mcp__flow-nexus__neural_predict({
  model_id: deployment.model_id,
  input: ["This product is amazing!", "Terrible experience."]
})

Time Series Forecasting

// Train LSTM model
const training = await mcp__flow-nexus__neural_train({
  config: {
    architecture: {
      type: "lstm",
      layers: [
        { type: "lstm", units: 128, return_sequences: true },
        { type: "dropout", rate: 0.2 },
        { type: "lstm", units: 64 },
        { type: "dense", units: 1 }
      ]
    },
    training: {
      epochs: 150,
      batch_size: 64,
      learning_rate: 0.01,
      optimizer: "adam"
    }
  },
  tier: "medium"
})

// Monitor progress
const status = await mcp__flow-nexus__neural_training_status({
  job_id: training.job_id
})

Federated Learning for Privacy

// Initialize federated cluster
const cluster = await mcp__flow-nexus__neural_cluster_init({
  name: "federated-medical-cluster",
  architecture: "transformer",
  topology: "mesh",
  consensus: "proof-of-learning",
  daaEnabled: true
})

// Deploy nodes across different locations
for (let i = 0; i < 5; i++) {
  await mcp__flow-nexus__neural_node_deploy({
    cluster_id: cluster.cluster_id,
    node_type: "worker",
    model: "large",
    autonomy: 0.9
  })
}

// Train with federated learning (data never leaves nodes)
await mcp__flow-nexus__neural_train_distributed({
  cluster_id: cluster.cluster_id,
  dataset: "medical_records_distributed",
  epochs: 200,
  federated: true,
  aggregation_rounds: 100
})

Architecture Patterns

Feedforward Networks

Best for: Classification, regression, simple pattern recognition

{
  type: "feedforward",
  layers: [
    { type: "dense", units: 256, activation: "relu" },
    { type: "dropout", rate: 0.3 },
    { type: "dense", units: 128, activation: "relu" },
    { type: "dense", units: 10, activation: "softmax" }
  ]
}

LSTM Networks

Best for: Time series, sequences, forecasting

{
  type: "lstm",
  layers: [
    { type: "lstm", units: 128, return_sequences: true },
    { type: "lstm", units: 64 },
    { type: "dense", units: 1 }
  ]
}

Transformers

Best for: NLP, attention mechanisms, large-scale text

{
  type: "transformer",
  layers: [
    { type: "embedding", vocab_size: 10000, embedding_dim: 512 },
    { type: "transformer_encoder", num_heads: 8, ff_dim: 2048 },
    { type: "global_average_pooling" },
    { type: "dense", units: 2, activation: "softmax" }
  ]
}

GANs

Best for: Generative tasks, image synthesis

{
  type: "gan",
  generator_layers: [...],
  discriminator_layers: [...]
}

Autoencoders

Best for: Dimensionality reduction, anomaly detection

{
  type: "autoencoder",
  encoder_layers: [
    { type: "dense", units: 128, activation: "relu" },
    { type: "dense", units: 64, activation: "relu" }
  ],
  decoder_layers: [
    { type: "dense", units: 128, activation: "relu" },
    { type: "dense", units: input_dim, activation: "sigmoid" }
  ]
}

Best Practices

1. Start Small: Begin with nano or mini tiers for experimentation
2. Use Templates: Leverage marketplace templates for common tasks
3. Monitor Training: Check status regularly to catch issues early
4. Benchmark Models: Always benchmark before production deployment
5. Distributed Training: Use clusters for large models (>1B parameters)
6. Federated Learning: Use for privacy-sensitive data
7. Version Models: Publish successful models as templates for reuse
8. Validate Thoroughly: Use validation workflows before deployment

Troubleshooting

Training Stalled

// Check cluster status
const status = await mcp__flow-nexus__neural_cluster_status({
  cluster_id: "cluster_id"
})

// Terminate and restart if needed
await mcp__flow-nexus__neural_cluster_terminate({
  cluster_id: "cluster_id"
})

Low Accuracy

• Increase epochs
• Adjust learning rate
• Add regularization (dropout)
• Try different optimizer
• Use data augmentation

Out of Memory

• Reduce batch size
• Use smaller model tier
• Enable gradient accumulation
• Use distributed training

Related Skills

• flow-nexus-sandbox - E2B sandbox management
• flow-nexus-swarm - AI swarm orchestration
• flow-nexus-workflow - Workflow automation

Resources

• Flow Nexus Docs: https://flow-nexus.ruv.io/docs
• Neural Network Guide: https://flow-nexus.ruv.io/docs/neural
• Template Marketplace: https://flow-nexus.ruv.io/templates
• API Reference: https://flow-nexus.ruv.io/api

---

Note: Distributed training requires authentication. Register at https://flow-nexus.ruv.io or use npx flow-nexus@latest register.

Core Principles

Flow Nexus Neural Networks operates on 3 fundamental principles:

Principle 1: Distributed Training Through E2B Sandbox Orchestration

Scale neural network training across multiple isolated sandboxes with coordinated gradient aggregation and model synchronization.

In practice:

• Initialize cluster with topology (mesh, ring, hierarchical) for node communication patterns
• Deploy worker nodes (training), parameter servers (gradient aggregation), and aggregators (model sync) as separate sandboxes
• WASM-accelerated inference provides 10-100x faster training vs pure JavaScript implementations

Principle 2: Federated Learning Enables Privacy-Preserving Distributed Training

Train models on distributed data without centralizing sensitive information by keeping data on local nodes.

In practice:

• Worker nodes train on local data (medical records, user activity) without uploading raw data
• Aggregation rounds collect only model updates (gradients) from min 5 nodes per round
• Proof-of-Learning consensus validates training progress without exposing training data

Principle 3: Template Marketplace Accelerates Development Through Reusable Models

Leverage pre-trained models and proven architectures from community templates instead of building from scratch.

In practice:

• Deploy sentiment analysis, image classification, time series forecasting with single command
• Customize templates with training config overrides (epochs, learning rate, batch size)
• Publish successful models as templates for credits or free community contribution

Common Anti-Patterns

Anti-Pattern	Problem	Solution
Deploying All Workers Before Cluster Initialization	Workers deployed without cluster context fail to connect or coordinate	Initialize cluster first with neural_cluster_init, capture cluster_id, then deploy nodes with cluster_id reference
Ignoring Training Tiers	Using large tier for prototyping wastes compute; nano for production underfits	Match tier to task: nano/mini for experimentation, small/medium for production, large/xl for complex models (>1B parameters)
Sequential Node Deployment	Deploying 10 workers sequentially takes 50s (5s per sandbox); delays training start	Deploy nodes in parallel by calling neural_node_deploy concurrently for all workers (5s total vs 50s sequential)

Conclusion

Flow Nexus Neural Networks transforms distributed AI development by providing E2B sandbox infrastructure for scalable neural network training, federated learning for privacy-preserving model development, and a marketplace of reusable templates to accelerate common tasks. By orchestrating multiple sandboxes with coordinated gradient aggregation and model synchronization, you achieve enterprise-scale training capabilities without managing infrastructure.

Use this skill when training large models requiring distributed computing (>1B parameters across multiple GPUs), implementing privacy-sensitive ML applications (healthcare, finance with federated learning), or accelerating development with pre-trained templates (sentiment analysis, image classification, forecasting). The key insight is sandbox isolation combined with coordination: each node operates independently in secure E2B environment while participating in collective training through gradient sharing and model updates. Start with single-node training using templates for common tasks, scale to distributed clusters when model size or training time exceeds single-machine limits, and enable federated learning only when data cannot be centralized due to privacy or compliance constraints.