---
language:
- en
pipeline_tag: text-classification
library_name: transformers
tags:
- cybersecurity
- xss
- security
- web
- payload-detection
- text-classification
- web-security
---
# XSS Payload Detector

DistilBERT-based machine learning model for detecting XSS payloads.
This project can be used either as a standalone CLI classifier or as a log-monitoring system that automatically analyzes web server requests and detects potential XSS attacks.

## Labels

| Label | Description |
|---------|-------------|
| NORMAL | Benign input |
| XSS | Potential XSS payload |

## Requirements

```bash
pip install torch
pip install transformers
pip install flask
```

## Features

This project supports two different usage modes.

### 1. CLI Mode

Run:

```bash
python inference_bert_url.py
```

Enter a string or URL directly from the command line.

The model will classify the input as:

- NORMAL
- XSS

and display a confidence score.

Example:

```text
Input:
<script>alert(1)</script>

Result:
XSS
Confidence:
0.9998
```

---

### 2. Log Monitoring Mode

Run the monitoring service:

```bash
python monitor.py
```

The monitor automatically reads web server access logs and analyzes incoming requests.

Detected XSS payloads are logged for further inspection.

This allows the model to be integrated into a web application environment without manually entering payloads.

---

### Test Environment

Run the example Flask server:

```bash
python test_server.py
```
![Demo](photos/xsstest.png)
Then open:

```text
http://127.0.0.1:8080/?q=abcde
```

Example XSS payload:

```text
http://127.0.0.1:8080/?q=<img src='x' onerror='alert("xss")'>
```
![Demo](photos/xssdetect.png)


![Demo](photos/pagetest.png)
The request will be reflected by the test page and analyzed by the monitoring service.

### Components

- `inference_bert_url.py`
  - Standalone CLI tool for testing XSS detection.

- `moniter.py`
  - Log monitoring service that reads web server logs and analyzes incoming requests.

- `test_server.py`
  - Flask-based demonstration server for testing reflected XSS scenarios.

- `templates/`
  - HTML templates used by the Flask demonstration server.

- `static/`
  - Static assets (images, CSS, etc.) used by the Flask demonstration server.

## Model Files

- config.json
- model.safetensors
- tokenizer.json
- tokenizer_config.json
- vocab.txt

## Download Repository

To download the entire repository, including the trained model, example server, monitoring utility, templates, and static files:

```bash
pip install huggingface_hub
```

```python
from huggingface_hub import snapshot_download

snapshot_download(
    repo_id="kd7979148/XSS_Payload_Detector"
)
```
This will download all files contained in the repository.

## Framework

- PyTorch
- Transformers