Jina AI
Jina AI provides web search and content extraction APIs. Reaktors use it to search the web and scrape individual pages at runtime, giving AI models access to up-to-date information beyond their training data.
You configure Jina as a Connection of type jina under Connections in the sidebar.
What Operaide Uses
Operaide uses two of Jina's API products through a single connection:
| Product | Endpoint | Purpose |
|---|---|---|
| Search | s.jina.ai | Web search — returns structured results for a query |
| Reader | r.jina.ai | Page scraping — extracts content from a URL as Markdown |
The platform derives the Reader URL from the configured base URL automatically.
Jina also offers Embeddings and Reranker APIs. These are separate products and are not used by this connection. Operaide handles embeddings through the AI Provider system.
No model selection is required. The Search and Reader APIs do not use model parameters.
Creating the Connection
- Navigate to Connections and click + Add Connection.
- Select Jina AI API as the connection type.
- Enter a connection name, e.g.
jina. - Fill in your API key (obtain one at jina.ai) and keep the default base URL (
https://s.jina.ai/). - Click Create.
Connection Fields
| Field | Required | Default | Description |
|---|---|---|---|
| API Key | Yes | — | Your Jina API key. Free keys are available at jina.ai. |
| Base URL | No | https://s.jina.ai/ | The Jina Search API endpoint. Only change this if you use a self-hosted or proxied Jina deployment. |
Common Mistakes
- Confusing Jina products. The Jina connection only covers Search and Reader. If you need embeddings, configure an embedding model through the AI Provider — not through this connection.
- Changing the base URL without understanding the derivation. The Reader URL is derived by replacing
s.jina.aiwithr.jina.aiin the base URL. If you set a custom base URL that does not follow this pattern, scraping will fail.