Implementation:Guardrails ai Guardrails RC

Knowledge Sources	Guardrails
Domains	Configuration, Settings
Last Updated	2026-02-14 00:00 GMT

Overview

The RC dataclass represents the Guardrails runtime configuration loaded from the ~/.guardrailsrc file.

Description

This module defines the RC dataclass which reads and parses the user's .guardrailsrc configuration file from their home directory. The file uses a simple key=value format, one entry per line. The class handles:

• Boolean config parsing: Keys listed in BOOL_CONFIGS (no_metrics, enable_metrics, use_remote_inferencing) are automatically converted to boolean values.
• Quote stripping: Values wrapped in matching single or double quotes have the quotes removed.
• Legacy support: The deprecated no_metrics key is backfilled into enable_metrics with inverted logic for backwards compatibility.
• Graceful fallback: If the .guardrailsrc file does not exist, an empty default RC instance is returned.

The RC class extends Serializeable, providing from_dict deserialization support.

Usage

Use RC to access the user's Guardrails configuration. Call RC.load() to read and parse the configuration file, or RC.exists() to check if a configuration file is present. This is typically used internally by the framework to determine metrics and inferencing settings.

Code Reference

Source Location

• Repository: Guardrails
• File: guardrails/classes/rc.py
• Lines: 1-80

Signature

BOOL_CONFIGS = set(["no_metrics", "enable_metrics", "use_remote_inferencing"])

@dataclass
class RC(Serializeable):
    id: Optional[str] = None
    token: Optional[str] = None
    enable_metrics: Optional[bool] = True
    use_remote_inferencing: Optional[bool] = True

    @staticmethod
    def exists() -> bool:

    @classmethod
    def load(cls, logger: Optional[logging.Logger] = None) -> "RC":

Import

from guardrails.classes.rc import RC

I/O Contract

RC Fields

Field	Type	Default	Description
id	Optional[str]	None	User or instance identifier.
token	Optional[str]	None	Authentication token for Guardrails services.
enable_metrics	Optional[bool]	True	Whether metrics collection is enabled.
use_remote_inferencing	Optional[bool]	True	Whether to use remote inferencing for validators.

exists

Return Type	Description
bool	True if ~/.guardrailsrc exists on disk; False otherwise.

load

Parameter	Type	Default	Description
logger	Optional[logging.Logger]	None	Logger instance for warning on malformed lines. Defaults to the root logger.

Return Type	Description
RC	A populated RC instance. Returns a default instance with empty values if the file is not found.

Configuration File Format

The ~/.guardrailsrc file uses a simple key-value format:

id=my-user-id
token=my-secret-token
enable_metrics=true
use_remote_inferencing=false

Usage Examples

from guardrails.classes.rc import RC

# Check if configuration exists
if RC.exists():
    config = RC.load()
    print(f"Token: {config.token}")
    print(f"Metrics enabled: {config.enable_metrics}")
    print(f"Remote inferencing: {config.use_remote_inferencing}")
else:
    print("No .guardrailsrc found, using defaults")
    config = RC()  # Uses default values

# Load with a custom logger
import logging
logger = logging.getLogger("my_app")
config = RC.load(logger=logger)

Related Pages

• Guardrails_ai_Guardrails_CLI_Validate - CLI command that may use RC settings for validation configuration
• Guardrails_ai_Guardrails_CLI_Watch - CLI command that reads settings influenced by the RC configuration