Integration Guide
agentAssert for AutoGen
Behavioral contracts for conversational agent patterns.
Why agentAssert + AutoGen
Conversation Governance
Enforce contracts on multi-turn agent conversations. Monitor message flows, track state across turns, and bound divergent discussions.
Azure AI Compatible
Works with Azure AI Foundry deployments. Deploy governed agents on Microsoft infrastructure with no additional configuration.
GroupChat Enforcement
Govern multi-agent GroupChat sessions. Contracts enforce speaker selection policies, turn limits, and topic adherence across all participants.
Get Started
Install, wrap your agent, and enforce.
# Request evaluation access at /contact
# Integration packages provided with accessfrom agentassert.integrations import AutoGenEnforcer
# Load contract and wrap your agent
enforcer = AutoGenEnforcer.from_file("contract.yaml")
enforced_agent = enforcer.wrap(your_agent)Frequently Asked Questions
How does agentAssert govern GroupChat sessions?
agentAssert monitors the entire GroupChat message stream. Contracts can enforce speaker rotation policies, content boundaries, and cross-agent consistency. Violations trigger configurable recovery actions.
Is it compatible with Azure AI Foundry?
Yes. agentAssert is a Python library that wraps your AutoGen agents. It works with any model backend, including Azure OpenAI Service and Azure AI Foundry deployments.
Can I govern nested conversations?
Yes. When an agent initiates a nested conversation, the parent contract scope extends to the nested context. You can also assign separate contracts to nested conversations.
Does it support human-in-the-loop patterns?
Fully supported. Human proxy agents participate in the governed conversation like any other agent. Contracts can define specific rules for human intervention points.
What AutoGen versions are supported?
agentAssert supports AutoGen 0.4+ and the latest AutoGen Studio. The integration adapter handles version-specific differences automatically.