← Eric Yocam's projects
Repo

Multi-Agent Steganographic Prompt Injection Defense

@ericyocamby Eric Yocam · California Polytechnic State University

Proof-of-concept that shows how a malicious instruction hidden in an image (LSB steganography) can be extracted and executed by a multi-agent system. Uses a Coordinator + 4 Workers + Verifier architecture over real MCP tools, with a local Qwen model driving agent decisions. Demonstrates a vulnerable path (stego-extracted text treated as trusted) and a protected path (source tagging + policy blocking of critical tools). Includes before/after visualization of the cover image and an email-attachment attack scenario.

Built withPythonGoogle ColabMCP (Model Context Protocol)Qwen2.5-0.5B-InstructHugging Face TransformersLSB SteganographyPillowNumPymulti-agentMCP steganographyprompt-injectionAI-securityagentic-AI QwenHugging-FaceColabLSB

Comments · 0

No comments yet. Tried it? Tell Eric what worked, what didn't, or what you'd want next.
Sign in to comment

More from members

All projects

See what other members are building and what each project is looking for.