v1.1.0 - RAG & Multi-Agent Security

What's New in v1.1.0

This release introduces comprehensive security scanning for RAG (Retrieval-Augmented Generation) systems and Multi-Agent orchestration frameworks.

RAG Security Scanner

• 30+ security rules covering vector stores, document ingestion, and poisoning detection
• 8 vector stores supported: Pinecone, Chroma, Weaviate, Qdrant, Milvus, PGVector, Redis, FAISS
• New CLI commands: secureagent rag scan, rag vector-stores, rag documents, rag poisoning, rag test

Multi-Agent Security Scanner

• 30+ security rules for orchestration, communication, and delegation attacks
• 4 frameworks supported: LangGraph, AutoGen, CrewAI, AutoGPT
• Detects: workflow cycles, privilege escalation, circular delegation, message injection, replay attacks
• New CLI commands: secureagent multiagent scan, multiagent orchestration, multiagent communication, multiagent delegation, multiagent frameworks

Jailbreak Detection

• New secureagent detect command group for jailbreak and prompt injection detection

Security Testing Payloads

• Multi-agent attack payloads (MA-001 to MA-010)
• RAG poisoning payloads

Test Coverage

• 308 tests passing
• Comprehensive test coverage for all new scanners

---

Installation

pip install secureagent

Quick Start

# Scan RAG system
secureagent rag scan ./rag-project

# Scan multi-agent system
secureagent multiagent scan ./agents

# Detect jailbreak attempts
secureagent detect jailbreak "user input"

---

Full Changelog: v1.0.2...v1.1.0