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

# Reference Check

> Submit a document and have its references evaluated and linked using Scite's data.

Reference Check takes a document (PDF or DOCX), extracts its references, and evaluates each one against Scite's citation data. It surfaces retractions, contradicting citations, and other flags associated with the cited sources. This is an asynchronous, poll-based flow.

**Need an API key?** On Pro, create one in the [API Console](https://scite.ai/users/me/api). Creating a key doesn't grant every optional scope; see [Authentication](/authentication) for feature access. For higher limits or managed credentials, [contact sales](https://scite.ai/contact).

<Warning>
  Reference Check requires a paid license and isn't included in a standard self-service Pro key. [Email sales](mailto:sales@scite.ai) to get access, then generate a new key from the [API Console](https://scite.ai/users/me/api) once it's enabled on your account.
</Warning>

## 1. Submit the document

You can either upload the file directly or point to a URL:

```bash theme={null}
# Upload directly
curl -X POST 'https://api.scite.ai/reference_check' \
  -H 'Authorization: Bearer <YOUR_API_KEY>' \
  -F 'file=@paper.pdf'
```

```bash theme={null}
# Or reference a URL
curl -X POST 'https://api.scite.ai/reference_check' \
  -H 'Authorization: Bearer <YOUR_API_KEY>' \
  -H 'Content-Type: application/json' \
  -d '{"url": "https://example.com/paper.pdf"}'
```

The response contains a task ID:

```json theme={null}
{ "id": "task_123" }
```

## 2. Poll for the result

```bash theme={null}
curl 'https://api.scite.ai/reference_check/tasks/task_123' \
  -H 'Authorization: Bearer <YOUR_API_KEY>'
```

Poll until the task status is no longer pending. The response includes the current state and, once complete, the evaluated references.

Back off between polling requests and enforce an overall timeout. See [Errors and rate limits](/errors-and-rate-limits#polling-asynchronous-tasks).

<Note>
  An unrecognized or mistyped `task_id` doesn't return a 404, it returns `200 {"status": "PENDING"}` just like a real in-progress task. Hold onto the exact ID from the submit response rather than reconstructing it, or you can end up polling a task that will never complete.
</Note>

## Other task operations

| Endpoint                                          | Purpose                                                       |
| ------------------------------------------------- | ------------------------------------------------------------- |
| `GET /reference_check/tasks/{task_id}/cancel`     | Cancel a running task                                         |
| `GET /reference_check/tasks/{task_id}/result_url` | Get a direct URL to the result, instead of the inline payload |

## Related

* [Assistant guide](/guides/assistant): a similar poll-based flow for Q\&A instead of reference evaluation
* [Errors and rate limits](/errors-and-rate-limits): retry and polling guidance