Implementation:Protectai Llm guard Output BanCode

Knowledge Sources	Protectai_Llm_guard
Domains	Security, Code_Detection
Last Updated	2026-02-14 12:00 GMT

Overview

BanCode is an output scanner that detects and blocks code snippets in LLM responses by delegating to the input-side InputBanCode scanner.

Description

The BanCode output scanner is a thin wrapper around the corresponding input scanner InputBanCode. It intercepts LLM outputs and checks whether they contain code snippets that should be blocked. Internally, the scanner creates an instance of InputBanCode with the same configuration parameters and delegates the scanning of the output text to it. This design pattern keeps the code DRY while providing a consistent output scanner interface. The scanner uses a classification model to detect code presence and compares against a configurable threshold to determine whether the output should be flagged.

Usage

Use this scanner when you need to prevent an LLM from returning code in its responses. This is useful in scenarios where code generation is not desired, such as customer support chatbots, educational tools that should explain concepts without providing ready-made code, or compliance-sensitive applications where code output could pose a security risk.

Code Reference

Source Location

• Repository: Protectai_Llm_guard
• File: llm_guard/output_scanners/ban_code.py
• Lines: 1-37

Signature

class BanCode(Scanner):
    def __init__(
        self,
        *,
        model: Model | None = None,
        threshold: float = 0.9,
        use_onnx: bool = False,
    ) -> None: ...

    def scan(self, prompt: str, output: str) -> tuple[str, bool, float]: ...

Import

from llm_guard.output_scanners import BanCode

I/O Contract

Inputs

Name	Type	Required	Description
prompt	str	Yes	The input prompt
output	str	Yes	The LLM output to scan for code snippets

Constructor Parameters

Name	Type	Required	Default	Description
model	None	No	None	Custom classification model to use for code detection
threshold	float	No	0.9	Confidence threshold above which code is flagged
use_onnx	bool	No	False	Whether to use ONNX runtime for inference

Outputs

Name	Type	Description
sanitized_output	str	The output (potentially modified)
is_valid	bool	Whether the output passed the scan (True if no code detected)
risk_score	float	Risk score (-1.0 to 1.0)

Usage Examples

Basic Usage

from llm_guard.output_scanners import BanCode

scanner = BanCode(threshold=0.9)

prompt = "Explain how sorting works"
output = "Sorting arranges elements in order. Here is an example:\ndef bubble_sort(arr):\n    for i in range(len(arr)):\n        for j in range(len(arr)-1):\n            if arr[j] > arr[j+1]:\n                arr[j], arr[j+1] = arr[j+1], arr[j]"

sanitized_output, is_valid, risk_score = scanner.scan(prompt, output)

if not is_valid:
    print(f"Code detected in output (risk score: {risk_score})")

Related Pages

• Principle:Protectai_Llm_guard_Code_Snippet_Detection