Skip to content

Configuration

Jump to bottom
Michael Goetz edited this page Jun 14, 2025 · 3 revisions

Configuration Guide

This guide covers all configuration options for the Multi-Agent Framework.

Table of Contents

Configuration Files

File Locations

your-project/
├── .maf-config.json      # Main configuration
├── .env                  # API keys and secrets
└── .maf/                 # Runtime directory (auto-generated)
    ├── state.json        # Framework state tracking
    └── message_queues/   # Agent message queues

Configuration Hierarchy

  1. Command-line arguments (highest priority)
  2. Environment variables
  3. .maf-config.json file
  4. Default values (lowest priority)

Project Configuration

Basic Structure

{
  "project_name": "My Awesome App",
  "project_type": "nextjs",
  "description": "A Next.js application with AI agents",
  "version": "1.0.0",
  "framework_config": {
    "state_file": ".maf/state.json",
    "message_queue_dir": ".maf/message_queues",
    "log_dir": ".maf_logs",
    "temp_dir": ".maf_temp"
  }
}

Project Types

Type Description Auto-Detection
nextjs Next.js applications next.config.js, app/ or pages/
react React applications package.json with react
django Django projects manage.py, settings.py
fastapi FastAPI applications FastAPI imports
express Express.js apps Express imports
auto Auto-detect type Default

Technology Stack

{
  "technology_stack": {
    "frontend": ["react", "typescript", "tailwind"],
    "backend": ["fastapi", "postgresql"],
    "testing": ["pytest", "jest"],
    "deployment": ["docker", "kubernetes"]
  }
}

Agent Configuration

Enabling/Disabling Agents

{
  "agent_config": {
    "enabled_agents": [
      "orchestrator",      // Always required
      "frontend_agent",
      "backend_agent",
      "db_agent",
      "qa_agent",
      "devops_agent",      // NEW: DevOps & Infrastructure
      "security_agent",    // NEW: Security Auditing
      "docs_agent",        // NEW: Documentation
      "ux_ui_agent"        // NEW: Design Systems
      // Omit agents you don't need
    ]
  }
}

Agent-Specific Settings

{
  "agent_config": {
    "agent_settings": {
      "frontend_agent": {
        "framework_preference": "react",
        "styling": "tailwind",
        "typescript": true
      },
      "backend_agent": {
        "framework": "fastapi",
        "database": "postgresql",
        "auth_method": "jwt"
      },
      "qa_agent": {
        "test_framework": "pytest",
        "coverage_threshold": 80,
        "strict_mode": true
      }
    }
  }
}

Task Assignment Rules

{
  "agent_config": {
    "task_routing": {
      "ui_tasks": ["frontend_agent", "ux_agent"],
      "api_tasks": ["backend_agent"],
      "data_tasks": ["db_agent", "backend_agent"],
      "test_tasks": ["qa_agent"],
      "deploy_tasks": ["devops_agent"]
    }
  }
}

Model Providers

Provider Configuration

{
  "agent_config": {
    "default_model_provider": "gemini",
    "default_model_name": "gemini-2.0-flash-exp",
    "providers": {
      "gemini": {
        "api_key_env": "GEMINI_API_KEY",
        "base_url": "https://generativelanguage.googleapis.com/v1beta"
      },
      "anthropic": {
        "api_key_env": "ANTHROPIC_API_KEY",
        "base_url": "https://api.anthropic.com"
      },
      "openai": {
        "api_key_env": "OPENAI_API_KEY",
        "base_url": "https://api.openai.com/v1"
      }
    }
  }
}

Per-Agent Models

{
  "agent_config": {
    "agent_models": {
      "orchestrator": {
        "provider": "anthropic",
        "name": "claude-3-sonnet-20240229",
        "temperature": 0.3,
        "max_tokens": 4096
      },
      "frontend_agent": {
        "provider": "openai",
        "name": "gpt-4-turbo-preview",
        "temperature": 0.7,
        "max_tokens": 8192
      },
      "qa_agent": {
        "provider": "gemini",
        "name": "gemini-2.0-flash-exp",
        "temperature": 0.1,
        "max_tokens": 4096
      }
    }
  }
}

Model Parameters

{
  "model_parameters": {
    "temperature": 0.7,        // Creativity (0.0-1.0)
    "max_tokens": 4096,        // Max response length
    "top_p": 0.9,             // Nucleus sampling
    "frequency_penalty": 0.0,  // Reduce repetition
    "presence_penalty": 0.0,   // Encourage new topics
    "stop_sequences": []       // Stop generation triggers
  }
}

Event Bus Configuration

In-Memory Event Bus (Default)

{
  "event_bus": {
    "type": "in_memory",
    "config": {
      "max_queue_size": 1000,
      "event_ttl": 3600,
      "cleanup_interval": 300
    }
  }
}

Kafka Event Bus (Scalable)

{
  "event_bus": {
    "type": "kafka",
    "config": {
      "bootstrap_servers": "localhost:9092",
      "topic_prefix": "maf_",
      "consumer_group": "maf_agents",
      "auto_offset_reset": "earliest",
      "enable_auto_commit": true
    }
  }
}

Event Filtering

{
  "event_bus": {
    "filters": {
      "log_events": ["task_created", "task_completed", "task_failed"],
      "persist_events": ["task_completed", "code_generated"],
      "priority_events": ["task_failed", "security_alert"]
    }
  }
}

Advanced Options

Performance Tuning

{
  "performance": {
    "parallel_agents": 5,           // Max concurrent agents
    "task_timeout": 300,            // Task timeout in seconds
    "retry_attempts": 3,            // Failed task retries
    "retry_delay": 60,              // Retry delay in seconds
    "batch_size": 10,               // Task batch size
    "cache_enabled": true,          // Enable response caching
    "cache_ttl": 3600              // Cache TTL in seconds
  }
}

Resource Limits

{
  "resource_limits": {
    "max_memory_mb": 2048,          // Per-agent memory limit
    "max_cpu_percent": 50,          // Per-agent CPU limit
    "max_file_size_mb": 100,        // Max generated file size
    "max_files_per_task": 20,       // Max files per task
    "rate_limit_per_minute": 60     // API rate limiting
  }
}

Security Options

{
  "security": {
    "enable_code_scanning": true,    // Scan generated code
    "allowed_imports": [             // Whitelist imports
      "react", "next", "express"
    ],
    "blocked_operations": [          // Dangerous operations
      "eval", "exec", "__import__"
    ],
    "sanitize_inputs": true,         // Input sanitization
    "audit_logging": true            // Security audit logs
  }
}

Integration Options

{
  "integrations": {
    "git": {
      "auto_commit": false,          // Auto-commit changes
      "branch_prefix": "maf/",       // Branch naming
      "commit_message_template": "feat: {task_description}"
    },
    "ci_cd": {
      "provider": "github_actions",  // CI/CD provider
      "auto_pr": false,              // Auto-create PRs
      "run_tests": true              // Run tests before PR
    },
    "monitoring": {
      "provider": "datadog",         // Monitoring service
      "metrics_enabled": true,       // Send metrics
      "traces_enabled": true         // Send traces
    }
  }
}

Environment Variables

Required Variables

# At least one LLM provider key
GEMINI_API_KEY=your_key_here
ANTHROPIC_API_KEY=your_key_here
OPENAI_API_KEY=your_key_here

Optional Variables

# Framework configuration
MAF_PROJECT_DIR=/path/to/project
MAF_CONFIG_FILE=custom-config.json
MAF_LOG_LEVEL=DEBUG
MAF_TIMEOUT=600

# Model configuration
MAF_DEFAULT_PROVIDER=gemini
MAF_DEFAULT_MODEL=gemini-2.0-flash-exp
MAF_TEMPERATURE=0.7

# Performance
MAF_PARALLEL_AGENTS=10
MAF_CACHE_ENABLED=true
MAF_RETRY_ATTEMPTS=5

# Security
MAF_ENABLE_SCANNING=true
MAF_AUDIT_LOGGING=true

Provider-Specific Variables

# Gemini
GOOGLE_API_KEY=your_key_here
GEMINI_PROJECT_ID=your_project_id

# Anthropic
ANTHROPIC_API_KEY=your_key_here
ANTHROPIC_API_VERSION=2023-06-01

# OpenAI
OPENAI_API_KEY=your_key_here
OPENAI_ORG_ID=your_org_id

Best Practices

1. API Key Security

# Never commit .env files
echo ".env" >> .gitignore

# Use separate keys for different environments
GEMINI_API_KEY_DEV=dev_key_here
GEMINI_API_KEY_PROD=prod_key_here

2. Model Selection

{
  "agent_models": {
    // Use stronger models for planning
    "orchestrator": {
      "provider": "anthropic",
      "name": "claude-3-opus-20240229"
    },
    // Use faster models for simple tasks
    "doc_agent": {
      "provider": "gemini",
      "name": "gemini-2.0-flash-exp"
    }
  }
}

3. Performance Optimization

{
  "performance": {
    // Adjust based on your system
    "parallel_agents": 3,      // Lower for limited resources
    "cache_enabled": true,     // Reduce API calls
    "batch_size": 5           // Balance speed vs resources
  }
}

4. Project-Specific Config

{
  // Next.js specific
  "agent_settings": {
    "frontend_agent": {
      "use_app_router": true,
      "server_components": true,
      "typescript_strict": true
    }
  }
}

5. Environment-Based Config

// config.js
const config = {
  development: {
    "log_level": "DEBUG",
    "retry_attempts": 1
  },
  production: {
    "log_level": "INFO",
    "retry_attempts": 3
  }
};

export default config[process.env.NODE_ENV || 'development'];

Configuration Examples

Minimal Configuration

{
  "project_name": "My App",
  "project_type": "auto"
}

Full-Stack Application

{
  "project_name": "E-Commerce Platform",
  "project_type": "nextjs",
  "technology_stack": {
    "frontend": ["nextjs", "typescript", "tailwind"],
    "backend": ["fastapi", "postgresql", "redis"],
    "testing": ["jest", "pytest", "playwright"]
  },
  "agent_config": {
    "enabled_agents": [
      "orchestrator",
      "frontend_agent",
      "backend_agent",
      "db_agent",
      "qa_agent",
      "security_agent"
    ],
    "agent_models": {
      "orchestrator": {
        "provider": "anthropic",
        "name": "claude-3-sonnet-20240229"
      },
      "frontend_agent": {
        "provider": "gemini",
        "name": "gemini-2.0-flash-exp"
      }
    }
  },
  "performance": {
    "parallel_agents": 5,
    "cache_enabled": true
  }
}

API-Only Project

{
  "project_name": "REST API Service",
  "project_type": "fastapi",
  "agent_config": {
    "enabled_agents": [
      "orchestrator",
      "backend_agent",
      "db_agent",
      "qa_agent",
      "doc_agent"
    ],
    "agent_settings": {
      "backend_agent": {
        "api_style": "rest",
        "auth": "oauth2",
        "database": "postgresql"
      }
    }
  }
}

Validation

Check Configuration

# Validate current config
maf config validate

# Test with specific config
maf config validate --config test-config.json

Common Validation Errors

  • Missing required fields
  • Invalid project type
  • Unsupported model providers
  • Conflicting settings
  • Invalid JSON syntax

Next Steps

Clone this wiki locally