This page may contain stale information. Last updated: 2026-06-03
Definition
Microsoft open-source framework (Adaptive Spec-driven Scoring for Evaluation and Regression Testing) converting natural-language policies into executable AI agent test cases.
Key Points
- MIT license; announced at Build 2026
- Integrates with LangChain, CrewAI, LiteLLM, OpenAI via OpenTelemetry
- Complements Agent Control Specification (ACS)
- Records tool calls and agent execution paths for regression testing