> ## 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. # Assertions > An assertion is a check that passes or fails a completed run. An assertion inspects a completed run and returns `null` to pass or a failure message to fail. Veval ships built-in assertions for common checks, and you can write your own by implementing `ITraceAssertion`. ## How assertions fit in You attach assertions to a [scenario](/concepts/scenarios), to a single scenario item, or to a [replay](/concepts/replay) through `ReplayOptions`. Failure messages are collected on the result, so a failing test tells you which check failed and why. ## Built-in assertions | Assertion | Fails when | | - | - | | `NoErrors` | Any step completed with an error status. | | `MaxSteps` | The total number of steps exceeds the limit. | | `MaxCost` | The total `cost_usd` across all steps exceeds the limit. | | `MaxDuration` | The total step duration exceeds the limit in milliseconds. | | `StepExists` | No step with the given name exists anywhere in the trace. | | `OutputContains` | No step's output contains the expected string. | | `ToolCalled` | No step with `type = "tool"` and the given name was recorded. | | `Judge` | An LLM judge, evaluated server-side, returns a failing verdict for the output. | | `MatchesSnapshot` | The run differs from a stored [snapshot](/concepts/snapshots), or the snapshot can't be loaded. | ## Deterministic checks and the judge Most assertions are deterministic. They read the recorded steps and need no LLM call. `Judge` grades the output against a rubric you write, using an LLM. The score and reasoning are recorded on the run whether it passes or fails. A judge call that fails outright counts as an assertion failure, never a silent pass. In replay, model outputs are recorded, so a code or prompt change shows up in step inputs, not outputs. To check whether a change made outputs better or worse, run the judge against live runs. ## Deep dives Use the built-in assertions, configure the judge, and write your own. Compare runs against a known-good baseline.