> ## Documentation Index
> Fetch the complete documentation index at: https://docs.valkyrie.vals.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Results and outputs

> Download scores, task artifacts, or a specific run path.

Valkyrie distinguishes evaluated results from the files produced by an agent.

<CodeGroup>
  ```bash CLI theme={null}
  valkyrie run results <run-id> --path ./results.json
  ```

  ```python Python SDK theme={null}
  results = await client.runs.results(run_id)
  ```
</CodeGroup>

Full syntax is in the [`run results`](/reference/cli/run#results), [`run outputs`](/reference/cli/run#outputs), and [`run output`](/reference/cli/run#output) reference, and in [`client.runs.results`](/reference/sdk/runs#results) for Python.

## Download results

Results hold the scores and evaluation. Without `--path`, the CLI writes `./results-<run-id>.json`.

Pass a task subset to recompute `final_score` over just those tasks. The stored full result is unchanged:

```bash theme={null}
valkyrie run results <run-id> --task-ids task_1,task_2
```

<Warning>
  Combining `--s3` with a task subset overwrites the canonical S3 result key with that subset view. Run the command again without filters to restore the full result.
</Warning>

## Download outputs

Outputs are the files the agent produced. Fetch all of them, or only the tasks you care about:

```bash theme={null}
valkyrie run outputs <run-id> --output-dir ./outputs
valkyrie run outputs <run-id> --task-ids astropy__astropy-7606,django__django-10880
```

To pull one file or directory instead of the whole run, name a path inside the run's benchmark prefix:

```bash theme={null}
valkyrie run output <run-id> <subpath> --output-dir ./output-dir
```

The output directory defaults to `./<run-id>`.