LLM-Based Format String Vulnerability Detection

This project implements a workflow for detecting format string vulnerabilities using Large Language Models (LLMs).

Configuration

1. Sync Dependencies: Run uv sync to synchronize project dependencies.

2. Install Required Tools: Download the required tools and place them in the tools/ directory. For detailed instructions, please refer to tools/README.md.

3. Configure LLM API Key: Set your LLM API key in the .env file. To use a different model, configure it in the config/ directory.

Quick Start

To start the vulnerability detection process, run:

uv run fmtvul-workflow --target-file <path-to-target-codebase-archive>

Replace <path-to-target-codebase-archive> with the path to your target codebase archive file.