Mirrors vs contract tests
Contract tests check shapes; a mirror checks behaviour across a whole conversation.
Where they differ
A contract test passes when the schema matches; trace conformance asks whether the agent would have done the same thing.
Both can run in the same pipeline; the mirror adds the behavioural gate.
When to pick which
contract tests remains the right tool when its own strengths are what you need; a mirror is for testing an agent against a backend that behaves like production without being production.
Related: /glossary/mirror/pricing