xpia

Here are 3 public repositories matching this topic...

freyzo / deep-xpia

Multi-hop cross-prompt injection benchmark for multi-agent AI systems. 250 attack cases, 8 taxonomy categories, 4 defenses evaluated. Watch: https://www.youtube.com/watch?v=fGOlMij4HPQ

benchmark mcp owasp multi-agent red-team ai-security prompt-injection agentic-ai llm-security-prompt-injection indirect-prompt-injection xpia

Updated Oct 11, 2026
Python

krishddd / SEC_Guardrails_Agent

Star

Defensive 7-layer runtime guardrails gateway for LLM agents: a reverse proxy that blocks prompt injection (incl. indirect/XPIA), tool misuse, and data exfiltration on every turn. Polyglot — Python control plane + Rust security core + React HITL. Metrics reported split (ASR vs FPR), never one blended F1.

python rust security reverse-proxy ai-safety guardrails llm prompt-injection llm-security agent-security prompt-injection-defense xpia

Updated Oct 18, 2026
Python

skajake1983 / xpia-tools-oss

Star

AI security research toolkit for cross-prompt injection (XPIA) testing — generate documents, payloads, QR codes, and web pages to evaluate LLM resilience. MIT-licensed.

react express typescript red-team security-research ai-security prompt-injection llm-security xpia

Updated Oct 17, 2026
TypeScript

Add this topic to your repo

To associate your repository with the xpia topic, visit your repo's landing page and select "manage topics."

Learn more