Ed1s0nZ/PrivHunterAI

What it solves

PrivHunterAI is a local permission detection workbench designed to identify authorization vulnerabilities. It automates the process of comparing HTTP(S) traffic from two different accounts (Account A and Account B) to see if one account can access resources it shouldn't be able to, effectively detecting broken access control issues.

How it works

The system acts as a passive proxy that captures HTTP(S) traffic. It automatically replaces credentials from Account A with those of Account B and replays the request to the same target. It then compares the responses. To analyze these results, it integrates with Chat Completions compatible LLMs to analyze the evidence and help determine if a vulnerability exists, while also allowing for manual review and confirmation by a security researcher.

Who it’s for

Security researchers and penetration testers who need to automate the detection of permission-related vulnerabilities in web applications.

Highlights

• Automated Credential Replacement: Automatically swaps credentials to test for unauthorized access.
• LLM-Assisted Analysis: Uses AI models to analyze response differences to help flag potential vulnerabilities.
• AI-Driven Evidence Desensitization: Automatically masks sensitive information in evidence before sending it to the LLM.
• Integrated Workbench: Provides a full UI for managing scans, reviewing findings, and exporting results in CSV or JSON formats.
• Local Persistence: Uses SQLite for storing users, settings, and audit logs.