Harness Configuration

Configuration File

The harness is configured via zod.config.js in your project root.

Basic Configuration

export default {
  harness: {
    maxAgents: 10,
    timeout: 30000,
    logLevel: 'info',
    executionMode: 'sync',
  },
  agents: {
    // Agent configurations
  },
};

Harness Options

maxAgents

• Type: number
• Default: 10
• Description: Maximum number of agents that can be registered

harness: {
  maxAgents: 50, // Allow up to 50 agents
}

timeout

• Type: number (milliseconds)
• Default: 30000
• Description: Default timeout for agent execution

harness: {
  timeout: 60000, // 60 second timeout
}

logLevel

• Type: string
• Default: 'info'
• Options: 'silent', 'error', 'warn', 'info', 'debug', 'trace'

harness: {
  logLevel: 'debug', // Verbose logging
}

executionMode

• Type: string
• Default: 'sync'
• Options: 'sync', 'async', 'stream'

harness: {
  executionMode: 'async', // Run agents in parallel
}

Agent Configuration

Define agents in the config file:

export default {
  agents: {
    assistant: {
      model: 'gpt-4',
      skills: ['conversation', 'reasoning'],
      tools: ['filesystem'],
      systemPrompt: 'You are a helpful assistant.',
      maxTokens: 2048,
      temperature: 0.7,
    },
    researcher: {
      model: 'gpt-4-turbo',
      skills: ['reasoning', 'web-search'],
      tools: ['web-search', 'database'],
      systemPrompt: 'You are a research specialist.',
    },
  },
};

Environment Variables

Configuration can be overridden via environment variables:

ZOD_HARNESS_MAX_AGENTS=20
ZOD_HARNESS_TIMEOUT=45000
ZOD_LOG_LEVEL=debug
ZOD_MODEL=gpt-4-turbo

Advanced Options

Retry Configuration

harness: {
  retry: {
    maxAttempts: 3,
    backoff: 'exponential', // or 'linear', 'fixed'
    initialDelay: 1000,
    maxDelay: 10000,
  },
}

Rate Limiting

harness: {
  rateLimit: {
    maxRequests: 100,
    perSeconds: 60,
    strategy: 'sliding-window', // or 'fixed-window', 'token-bucket'
  },
}

Persistence

harness: {
  persistence: {
    enabled: true,
    adapter: 'sqlite', // or 'postgres', 'redis', 'memory'
    connectionString: './data/messages.db',
    ttl: 86400, // 24 hours in seconds
  },
}

CORS (for HTTP API)

harness: {
  cors: {
    enabled: true,
    origin: ['https://example.com', 'http://localhost:3000'],
    methods: ['GET', 'POST'],
    allowedHeaders: ['Content-Type', 'Authorization'],
  },
}

Performance Tuning

Worker Threads

Enable worker threads for CPU-intensive tasks:

harness: {
  workers: {
    enabled: true,
    maxWorkers: 4,
    taskThreshold: 1000, // ms before offloading to worker
  },
}

Caching

Enable response caching:

harness: {
  cache: {
    enabled: true,
    adapter: 'redis', // or 'memory', 'memcached'
    ttl: 3600, // 1 hour
    maxSize: 1000, // max cache entries
  },
}

Production Checklist

• Set appropriate timeout values
• Configure logLevel to 'warn' or 'error' in production
• Enable persistence for message durability
• Set up rate limiting to prevent abuse
• Configure CORS if exposing HTTP API
• Enable caching for better performance
• Monitor resource usage with harness.getStats()

Example: Full Production Config

export default {
  harness: {
    maxAgents: 20,
    timeout: 45000,
    logLevel: 'warn',
    executionMode: 'async',
    retry: {
      maxAttempts: 3,
      backoff: 'exponential',
      initialDelay: 1000,
      maxDelay: 10000,
    },
    rateLimit: {
      maxRequests: 1000,
      perSeconds: 3600,
      strategy: 'sliding-window',
    },
    persistence: {
      enabled: true,
      adapter: 'postgres',
      connectionString: process.env.DATABASE_URL,
      ttl: 604800, // 7 days
    },
    cache: {
      enabled: true,
      adapter: 'redis',
      connectionString: process.env.REDIS_URL,
      ttl: 3600,
      maxSize: 5000,
    },
  },
  agents: {
    // Agent definitions
  },
};

Continue to Plugins to learn about extending the harness.