SecTrainingData

Data for training machine learning models related to bug bounty and pentesting.

Images Data

• Images of webpages organized by prediction tagging
• Images of VNC and RDP services (These will likely be two categories, normal ones with an expected login prompt and ones with no login or default username added)

Text Data

Nothing there at this time, but I have to have data for the following scenarios.

• Payloads for training attacks (SQLi, XSS, XXE, SSRF, Content brute forcing, Etc.)
• Tool output logs for anomaly detection

Images Usage Guide

1. Create project at https://www.customvision.ai

2. Upload Images using foldername as tag

3. Train your model

4. Once your model is trained you are all set to start sending images to your API.

Note: I'm currently working on a tool that will work with screenshotting tools to check against the API. For now the following bash script works ok.

IMAGES=$(ls | grep ".*\.png$")
URL="{Prediction URL}"
KEY="{Prediction Key"
for image in $IMAGES
do
  echo "Checking: $image"
  cat $image | curl -s -X POST "$URL" \
  -H "Prediction-Key: $KEY" \
  -H "Content-Type: application/octet-stream" \
  --data-binary "@-" | jq '.predictions[0] | {tagName,probability} '
done

5. After you have done some predictions. You can go back to the portal and add tags to ones you recently sent to the API make your predictions better.