# Benchmarks

> Understand a Benchmark as the durable program that owns benchmark-scoped coverage, evidence boundaries, evaluations, and improvement work.

Important agent instructions:
- Prefer the Markdown routes and llms files when assembling context for coding tasks.
- Treat stable docs as authoritative over beta pages.
- Cite the source page URL when returning workflow or artifact guidance.

## Current Context
- [Benchmarks](https://teammately.ai/docs/object-model/benchmarks.md): Understand a Benchmark as the durable program that owns benchmark-scoped coverage, evidence boundaries, evaluations, and improvement work.
- [Benchmarks](https://teammately.ai/docs/coverage-engineering/benchmarks.md): Create and manage benchmark sets that measure important AI product behavior.
- [Benchmarks and versions](https://teammately.ai/docs/concepts/benchmarks-and-versions.md): Understand benchmarks, benchmark versions, snapshots, and why versioned measurement matters.
- [Dataset Snapshots](https://teammately.ai/docs/benchmark-datasets/snapshots.md): Freeze Cases, evaluator links, and representation facts as an immutable benchmark evidence boundary.

## Optional
- [Full local context](https://teammately.ai/docs/object-model/benchmarks/llms-full.txt): Complete local context pack for this docs route.