Repository navigation
2026-06-25_AutoGen_Sovereign_Runtime
🚀 Release: AI SAFE² AutoGen 0.4 Sovereign Runtime
Protect AutoGen's CodeExecutorAgent before code ever executes.
The AI SAFE² Framework now includes the AutoGen 0.4 Sovereign Runtime, providing production-ready security enforcement for applications built on autogen_agentchat.
⚠️ Unlike other AI agent frameworks, AutoGen includes a CodeExecutorAgent capable of executing LLM-generated code. That creates a fundamentally different security problem where prompt injection can become arbitrary code execution.
✨ Highlights
🛡️ CodeBlockGuard pre-execution enforcement
⚡ Async-native security architecture
🔍 Prompt injection detection
🔑 Credential scanning
🌐 Network pattern restrictions
📜 Execution trace logging
📊 Compliance reporting
🏛️ ACT Tier 1–4 support
🚨 Mandatory CP.8 Catastrophic Risk Threshold enforcement for every CodeExecutorAgent
🧠 HEAR Doctrine enforcement for Class-H code patterns
🎯 Security Outcome
Instead of detecting malicious code after execution, the Sovereign Runtime evaluates every generated code block before it reaches the executor.
If the code violates policy, it never runs.
📚 Control Coverage
The runtime enforces AI SAFE² controls across:
- Prompt Security
- Semantic Isolation
- Memory Governance
- Cascade Containment
- Execution Trace Logging
- Tool Misuse Detection
- Compliance Validation
- Agentic Control Plane
- Catastrophic Risk Thresholds (CP.8)
📦 Get Started
Repository
https://github.com/CyberStrategyInstitute/ai-safe2-framework
AutoGen Sovereign Runtime
We welcome feedback, issues, and community contributions as AI SAFE² continues expanding production-ready security implementations for modern agentic AI frameworks.