MCP Config Validators

dnallm.mcp.config_validators

Configuration validators for MCP Server.

This module provides Pydantic models for validating MCP server configurations and inference model configurations.

Classes

InferenceConfig

Bases: BaseModel

Inference configuration for model prediction.

Methods:

set_use_bf16 classmethod

set_use_bf16(v, info)

Set use_bf16 based on precision setting.

Source code in dnallm/mcp/config_validators.py

@field_validator("use_bf16", mode="before")
@classmethod
def set_use_bf16(cls, v, info):
    """Set use_bf16 based on precision setting."""
    if info.data and "precision" in info.data:
        return info.data["precision"] == "bfloat16"
    return v

set_use_fp16 classmethod

set_use_fp16(v, info)

Set use_fp16 based on precision setting.

Source code in dnallm/mcp/config_validators.py

@field_validator("use_fp16", mode="before")
@classmethod
def set_use_fp16(cls, v, info):
    """Set use_fp16 based on precision setting."""
    if info.data and "precision" in info.data:
        return info.data["precision"] == "float16"
    return v

InferenceModelConfig

Bases: BaseModel

Complete inference model configuration.

Methods:

validate_task_labels classmethod

validate_task_labels(v)

Validate that num_labels matches label_names length.

Source code in dnallm/mcp/config_validators.py

@field_validator("task")
@classmethod
def validate_task_labels(cls, v):
    """Validate that num_labels matches label_names length."""
    if v.num_labels != len(v.label_names):
        raise ValueError("num_labels must match the length of label_names")
    return v

LoggingConfig

Bases: BaseModel

Logging configuration.

MCPConfig

Bases: BaseModel

MCP protocol configuration.

MCPServerConfig

Bases: BaseModel

Complete MCP server configuration.

Methods:

validate_model_names classmethod

validate_model_names(v)

Validate that model names are unique.

Source code in dnallm/mcp/config_validators.py

@field_validator("models")
@classmethod
def validate_model_names(cls, v):
    """Validate that model names are unique."""
    names = [entry.name for entry in v.values()]
    if len(names) != len(set(names)):
        raise ValueError("Model names must be unique")
    return v

validate_multi_model_references classmethod

validate_multi_model_references(v, info)

Validate that multi-model configurations reference existing models.

Source code in dnallm/mcp/config_validators.py

@field_validator("multi_model")
@classmethod
def validate_multi_model_references(cls, v, info):
    """Validate that multi-model configurations reference existing
    models."""
    if info.data and "models" in info.data:
        available_models = set(info.data["models"].keys())
        for config in v.values():
            for model_name in config.models:
                if model_name not in available_models:
                    raise ValueError(
                        f"Model '{model_name}' referenced in multi-model "
                        f"config but not defined in models"
                    )
    return v

warn_both_transports classmethod

warn_both_transports(v, info)

Warn when both SSE and Streamable HTTP blocks are present.

Having both sse and streamable_http configured is a valid transitional state, but it is unusual in production. A warning is logged so operators are aware that two transports are active.

Source code in dnallm/mcp/config_validators.py

@field_validator("streamable_http")
@classmethod
def warn_both_transports(cls, v, info):
    """Warn when both SSE and Streamable HTTP blocks are present.

    Having both ``sse`` and ``streamable_http`` configured is a valid
    transitional state, but it is unusual in production. A warning is
    logged so operators are aware that two transports are active.
    """
    import logging

    if v is not None and info.data and "sse" in info.data:
        logging.getLogger(__name__).warning(
            "Both 'sse' and 'streamable_http' blocks are present in the "
            "server configuration. This is valid for transitional "
            "deployments but unusual in production."
        )
    return v

ModelConfig

Bases: BaseModel

Individual model configuration.

ModelEntryConfig

Bases: BaseModel

Model entry in the main configuration.

Methods:

validate_config_path classmethod

validate_config_path(v)

Validate config path format.

Source code in dnallm/mcp/config_validators.py

@field_validator("config_path")
@classmethod
def validate_config_path(cls, v):
    """Validate config path format."""
    if not v or not v.strip():
        raise ValueError("Config path cannot be empty")
    return v

ModelInfoConfig

Bases: BaseModel

Model information configuration.

MultiModelConfig

Bases: BaseModel

Multi-model parallel prediction configuration.

SSEConfig

Bases: BaseModel

SSE (Server-Sent Events) configuration.

ServerConfig

Bases: BaseModel

Server configuration.

StreamableHTTPConfig

Bases: BaseModel

Streamable HTTP configuration.

Configuration for the Streamable HTTP transport protocol as defined in the MCP specification 2025-11-25. The default path of /mcp follows the standard MCP endpoint convention.

Attributes:

Name	Type	Description
host	str	Host address to bind the HTTP server to.
port	int	Port number to bind the HTTP server to.
path	str	URL path for the MCP endpoint (defaults to /mcp).

TaskConfig

Bases: BaseModel

Task configuration for DNA prediction models.

Functions:

validate_inference_model_config

validate_inference_model_config(config_path)

Validate inference model configuration file.

Source code in dnallm/mcp/config_validators.py

def validate_inference_model_config(config_path: str) -> InferenceModelConfig:
    """Validate inference model configuration file."""
    import yaml

    with open(config_path, encoding="utf-8") as f:
        config_dict = yaml.safe_load(f)

    return InferenceModelConfig(**config_dict)

validate_mcp_server_config

validate_mcp_server_config(config_path)

Validate MCP server configuration file.

Source code in dnallm/mcp/config_validators.py

def validate_mcp_server_config(config_path: str) -> MCPServerConfig:
    """Validate MCP server configuration file."""
    import yaml

    with open(config_path, encoding="utf-8") as f:
        config_dict = yaml.safe_load(f)

    return MCPServerConfig(**config_dict)