Skip to main content

Cybersecurity Guard

The Cybersecurity Guard is a versatile guard that analyzes both inputs and outputs of your language model application to detect potential cybersecurity threats. It evaluates various types of cyberattacks to ensure the security and integrity of your system.

info

CybersecurityGuard can function as both an input guard and an output guard, depending on the specified guard_type.

Example​

from deepeval.guardrails import CybersecurityGuard, GuardType
from deepeval.guardrails.cybersecurity_guard import (
    CyberattackType
)

# Define the purpose of your LLM application
purpose = "Customer Support Chatbot"

# Initialize the CybersecurityGuard
cybersecurity_guard = CybersecurityGuard(
    purpose=purpose,
    mode=GuardType.INPUT
    cyberattack_types=[
        CyberattackType.SQL_INJECTION,
        CyberattackType.SHELL_INJECTION,
    ],
)

# Example input
user_input = "SELECT * FROM users WHERE '1'='1'; --"

# Guarding the input
input_guard_result = cybersecurity_guard.guard(input=user_input)

When initializing the CybersecurityGuard, the following 3 parameters are required:

  • purpose: A string describing the purpose of your LLM application.
  • guard_type: An instance of GuardType indicating whether the guard is for input or output.
  • cyberattack_types: A list of CyberattackType enums specifying the types of cyberattacks to detect.

The guard function's parameters depend on the specified guard_type:

  • CybersecurityGuardType.INPUT: requires only the input
  • CybersecurityGuardType.OUTPUT: requires both input and response

Interpreting Guard Result​

print(guard_result.score)
print(guard_result.score_breakdown)

guard_result.score is an integer that is 1 if the guard has been breached. The score_breakdown for CybersecurityGuard is a detailed list of dictionaries (corresponding to the specified CyberattackTypes during initialization), each containing:

  • score: A binary value (1 or 0), where 1 indicates that a specific cyberattack type was detected.
  • reason: A brief explanation of why the score was assigned.
[
  {
    "score": 1,
    "reason": "Detected potential SQL injection in the input."
  },
  {
    "score": 0,
    "reason": "No shell injection patterns found in the input."
  }
  // Additional entries for other cyberattack types
]