# Ori Eval

**Tell your agent: run curl -fsSL https://openrouter.ai/skills/spawn-ori-eval and follow the instructions in its output to get started.**

If you are that agent: run the curl command above and follow its output. It is one
self-contained document that takes you from never having heard of Ori to a
finished eval, and it assumes no prior context.

## What Ori Eval is

Ori Eval answers "which model is best for what I'm building?" with evidence
instead of guesswork. An eval is a `*.eval.ts` file that runs your agent on your
own prompts, asserts on the tools it called, and grades open-ended answers
with an LLM judge.

Ori pins the harness and the model for the duration of a run,
so a score change means your agent changed, not the environment. Because it
routes through OpenRouter, a bakeoff can cover every major lab instead of one
vendor's models.

## Install

```sh
curl -fsSL https://openrouter.ai/labs/ori/install.sh | bash
```

Then run `ori login` to sign in. Running evals also needs
[Bun](https://bun.sh).

## Read more

- Skill for coding agents: https://openrouter.ai/skills/spawn-ori-eval
- Docs: https://openrouter.ai/docs/guides/ori/eval
- Human-facing page: https://openrouter.ai/ori/eval
- Every published skill: https://github.com/OpenRouterTeam/skills