# Multi-Chain Attacks

## Description
Targeting applications that chain multiple LLM calls that process and refine tasks sequentially.

## Attack Examples
- Feeding adversarial prompts to observe chain behavior
- Using tools like Garak and Giskard for testing
- Exploiting inter-model communication
- Manipulating sequential processing
- Creating chain reaction attacks
- Exploiting model handoff points
- Testing chain vulnerabilities
- Reference: https://labs.withsecure.com/publications/multi-chain-prompt-injection-attacks

## Tools
- https://github.com/NVIDIA/garak
- https://www.giskard.ai/