Skip to main content

What each integration evaluates

IntegrationStatusPromptsResponsesTool callsObservability
AWS Strands AgentsGenerally available✓✓✓
LangChainGenerally available✓✓✓
LangGraphGenerally available✓✓✓
ParlantGenerally available✓✓
Amazon Bedrock AgentCoreGenerally available✓
Amazon BedrockGenerally available✓
Microsoft Copilot StudioGenerally available✓✓
Microsoft Dynamics 365Generally available✓✓
LiteLLMGenerally available✓✓✓
Portkey AI GatewayBeta✓✓✓
Databricks Unity AI GatewayGenerally available✓✓✓
Kong AI GatewayBeta✓✓✓
Bifrost AI GatewayBeta✓✓✓
Databricks OmnigentPreview✓✓✓
Amazon Bedrock AgentCore GatewayPreview✓
Cognigy.AIComing soon✓✓
IBM MelleaGenerally available✓✓✓
Python SDKGenerally available✓✓
REST APIGenerally available✓✓
MLflowPreview✓✓
TypeScript SDKGenerally available✓✓

Microsoft Copilot Studio, Microsoft Dynamics 365 and Databricks Unity AI Gateway evaluate tool calls only once Alice has enabled WONDER_FENCE_AGENTIC_MESSAGE_TYPES on your project, against policies that list the Tool Call or Tool Result message type.

A2A (Agent2Agent) is not listed: it is a WonderBuild red-teaming target, and WonderFence evaluates nothing on it.