# Veval

## Docs

- [Assertions](https://docs.veval.dev/guides/assertions.md): Built-in checks that validate agent behavior after every run.
- [Replay](https://docs.veval.dev/guides/replay.md): Run your agent against recorded traces in unit and integration tests — no live LLM calls.
- [Scenarios](https://docs.veval.dev/guides/scenarios.md): Run your agent against multiple inputs and assert on every result.
- [Snapshots](https://docs.veval.dev/guides/snapshots.md): Detect structural regressions by comparing agent step shape against a pinned golden trace.
- [Tracing](https://docs.veval.dev/guides/tracing.md): Record every step of your agent and ship it to the Veval dashboard.
- [LLM Reference](https://docs.veval.dev/llm-reference.md): Complete Veval SDK reference for AI coding agents. Single-page context drop — all three languages, all APIs.
- [Quickstart](https://docs.veval.dev/quickstart.md): Start sending traces to Veval in minutes.

## Optional

- [Dashboard](https://dashboard.veval.dev)