> ## Documentation Index > Fetch the complete documentation index at: https://docs.veval.dev/llms.txt > Use this file to discover all available pages before exploring further. # Replay > Replay runs your agent against a recorded trace with mocked LLM responses. Replay runs your real agent code, but each step returns the output recorded in a [trace](/concepts/traces) instead of calling the LLM. Tests built on replay are deterministic, fast, and free. ## How replay fits in `VevalTestSdk` is a test double for `IVevalSdk`. Load a trace with `WithReplay`, inject the test SDK into your agent in place of the real one, and call it normally. When your agent calls `TrackStepAsync`, the test SDK returns the recorded output for that step name. Have your agent accept the `IVevalSdk` interface, not the concrete class, so your tests can swap in `VevalTestSdk`. Replay also powers trace-backed [scenario](/concepts/scenarios) items. ## Strict mocking `VevalTestSdk` never falls through to the live LLM. If your agent calls `TrackStepAsync` with a step name that isn't in the recorded trace, it throws: ```text theme={null} InvalidOperationException: Replay mode: no mock output for step 'new_step'. Available steps: classify, answer. This would have made a real LLM call. ``` A code change that adds a new LLM call fails the test instead of quietly making real calls. ## Guarding against drift Replay serves recorded outputs by step name. If your agent's control flow changes, a recorded output can be handed to the wrong call while every assertion still passes. Turn on `CompareWithRecording` in `ReplayOptions` and the replay also fails when its step sequence or inputs drift from the trace it's replaying. ## Reporting `VevalTestSdk` still reports traces and scenario results to the dashboard, so replay runs count in your history. ## Deep dives Set up `VevalTestSdk` in your test project. Compare replays against a known-good baseline.