Skip to content

Prototype AI agent to test how well AI can help us find and fix security problems

Notifications You must be signed in to change notification settings

gpsandhu23/ai_buddy_guard

Repository files navigation

AI Buddy Guard 🦾

Overview

Buddy Guard is a prototype AI agent to test how well AI can help us find and fix security problems.

Features

  • Secret Scanning: Finds any leaked crdentials in your code
  • Out of date dependencies: Finds any vulnerable dependencies in your git repo
  • Open S3 buckets: Checks for any open AWS S3 buckets
  • Missing MFA for AWS: Find users that don't have MFA enabled on AWS
  • Extract incident schema: Extract key insights about a security incident in a parsable schema from natural language text reports
  • Check CVE in KEV list: Check if a CVE is in the CISA KEV list
  • Threat model: Generate a basic threat model for a service given a url to documentation
  • Invalidate AWS Key: Invalite an AWS key if it has been compromised
  • Webpage phishing check: Check if a URL is a phishing webpage based on content, WHOIS records, TLS cert info and DNS records
  • Scan and generate code fixes: Scan code with GPT-4 and autogenerate code fixes. This is still very noisy, needs more work

Installation

Prerequisites

  • Docker
  • GitHub Personal Access Token for GitHub-related features
  • AWS tokens for AWS-related features

Steps

# Clone the repository
git clone https://github.com/gpsandhu23/ai_buddy_guard.git

# Navigate to the directory
cd aibuddyguard

# Build and run the Docker container
docker-compose up web_app

About

Prototype AI agent to test how well AI can help us find and fix security problems

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published