Configuration Options

MUXI Framework offers extensive configuration options to customize its behavior. This reference documents all available configuration parameters.

Configuration Format

MUXI supports configuration via:

• YAML files (default)
• JSON files
• Environment variables
• Direct programmatic configuration

The default configuration file is located at config.yaml in the project root. You can specify an alternative configuration file using the --config command-line option.

Core Configuration

Application

Parameter	Type	Default	Description
app.name	string	"MUXI Framework"	Application name
app.host	string	"127.0.0.1"	Host to bind the server to
app.port	integer	8000	Port to bind the server to
app.log_level	string	"info"	Logging level (debug, info, warning, error, critical)
app.environment	string	"development"	Environment (development, testing, production)
app.debug	boolean	false	Enable debug mode

Security

Parameter	Type	Default	Description
security.secret_key	string	N/A (Required)	Secret key for JWT tokens and encryption
security.token_expiry	integer	3600	JWT token expiry in seconds
security.cors_origins	list	["*"]	List of allowed CORS origins
security.cors_methods	list	["*"]	List of allowed CORS methods
security.cors_headers	list	["*"]	List of allowed CORS headers
security.rate_limit.enabled	boolean	true	Enable rate limiting
security.rate_limit.limit	integer	60	Default rate limit per minute

Database Configuration

PostgreSQL

Parameter	Type	Default	Description
database.url	string	"postgresql://postgres:postgres@localhost:5432/muxi"	Database connection URL
database.pool_size	integer	5	Connection pool size
database.max_overflow	integer	10	Maximum number of connections that can be created beyond pool_size
database.pool_recycle	integer	300	Connection recycle time in seconds
database.echo	boolean	false	Echo SQL queries for debugging

Memory Configuration

Parameter	Type	Default	Description
memory.buffer_size	integer	10	Size of the buffer memory (number of messages)
memory.search_k	integer	4	Number of memories to retrieve in semantic search
memory.similarity_threshold	float	0.7	Minimum similarity score for memory matches
memory.chunk_size	integer	1024	Size of text chunks for embeddings
memory.chunk_overlap	integer	200	Overlap between text chunks
memory.embedding_model	string	"text-embedding-ada-002"	Model to use for text embeddings

Models Configuration

OpenAI

Parameter	Type	Default	Description
models.openai.api_key	string	N/A (Required)	OpenAI API key
models.openai.organization	string	null	OpenAI organization ID
models.openai.default_model	string	"gpt-4"	Default model to use
models.openai.temperature	float	0.7	Temperature for sampling
models.openai.max_tokens	integer	1024	Maximum tokens in completion
models.openai.timeout	integer	120	API timeout in seconds
models.openai.retry_attempts	integer	3	Number of retry attempts for API calls

Anthropic

Parameter	Type	Default	Description
models.anthropic.api_key	string	N/A (Required)	Anthropic API key
models.anthropic.default_model	string	"claude-3-opus-20240229"	Default model to use
models.anthropic.temperature	float	0.7	Temperature for sampling
models.anthropic.max_tokens	integer	1024	Maximum tokens in completion
models.anthropic.timeout	integer	120	API timeout in seconds

MCP Configuration

Parameter	Type	Default	Description
mcp.enabled	boolean	true	Enable MCP system
mcp.servers	list	[]	List of MCP servers
mcp.discovery.enabled	boolean	true	Enable MCP server discovery
mcp.discovery.interval	integer	300	Server discovery interval in seconds
mcp.timeout	integer	30	Timeout for MCP server calls in seconds
mcp.retry_attempts	integer	2	Number of retry attempts for MCP calls

MCP Server Configuration Example

mcp:
  servers:
    - id: "weather"
      name: "Weather Service"
      url: "https://api.example.com/mcp/weather"
      auth:
        type: "api_key"
        key_name: "X-API-Key"
        key_value: "${WEATHER_API_KEY}"
      functions:
        - name: "getCurrentWeather"
          description: "Get current weather for a location"
          parameters:
            location:
              type: "string"
              description: "City name or coordinates"
            units:
              type: "string"
              enum: ["metric", "imperial"]
              default: "metric"

MCP Servers Configuration

Option	Type	Default	Description
mcp_servers.enabled	boolean	true	Enable MCP servers system
mcp_servers.search.url	string	"http://localhost:5001"	URL for the search MCP server
mcp_servers.search.api_key	string	null	API key for the search MCP server
mcp_servers.weather.url	string	"http://localhost:5002"	URL for the weather MCP server
mcp_servers.weather.api_key	string	null	API key for the weather MCP server
mcp_servers.calculator.url	string	"http://localhost:5003"	URL for the calculator MCP server

WebSocket Configuration

Parameter	Type	Default	Description
websocket.enabled	boolean	true	Enable WebSocket interface
websocket.ping_interval	integer	30	WebSocket ping interval in seconds
websocket.ping_timeout	integer	10	WebSocket ping timeout in seconds
websocket.max_message_size	integer	1048576	Maximum WebSocket message size in bytes

Environment Variables

All configuration options can be set via environment variables using the following format:

MUXI_[SECTION]_[PARAMETER]

For example:

• MUXI_APP_PORT=9000
• MUXI_SECURITY_SECRET_KEY=mysecretkey
• MUXI_MODELS_OPENAI_API_KEY=sk-...

Nested parameters use underscore as separator:

• MUXI_SECURITY_RATE_LIMIT_ENABLED=false

Programmatic Configuration

When using MUXI programmatically, you can set configuration options during initialization:

from muxi.core import MuxiApp

app = MuxiApp(
    config={
        "app": {
            "name": "My Custom MUXI App",
            "port": 9000
        },
        "models": {
            "openai": {
                "api_key": "sk-...",
                "default_model": "gpt-4"
            }
        }
    }
)

Configuration Precedence

Configuration values are loaded in the following order, with later sources overriding earlier ones:

1. Default values
2. Configuration file
3. Environment variables
4. Programmatic configuration

Sensitive Configuration

For security reasons, sensitive values like API keys should be provided through environment variables rather than in configuration files, especially in production environments.

When using a configuration file, you can reference environment variables using ${ENV_VAR} syntax:

models:
  openai:
    api_key: "${OPENAI_API_KEY}"

Validation

MUXI validates configuration options at startup and will raise errors for required values that are missing or invalid. If you see configuration validation errors, check that:

1. All required values are provided
2. Values are of the correct type
3. Enum values are from the allowed set
4. Referenced files or directories exist and are readable