Search the Web
jina.search
ActionAI & modelsconnection required
Calls the Jina AI API with the connection's credentials.
Run a web search and return the results already read into LLM-ready content, in search-engine results page order.
Connection
- Required
- Yes — the node will not save without one
- Auth type
- Bearer token
- Providers
- Jina AI
- Secret shape
- the API key, or {"api_key": "…"}
- Config it needs
- base_url, default_embedding_model, default_reranker_model
- Resolved by
- config.connection_id, else node.connection_id
Configuration
17 fields, 1 required. A dotted path is a nested object and name[] an array of objects — the editor renders the same tree.
| Field | Type | Required | Default | Constraints | Example |
|---|---|---|---|---|---|
query What to search for. | string | Yes | — | — | jina embeddings pricing |
site Restrict the search to one domain, for example https://jina.ai. | string | No | — | — | — |
country Two-letter country code the search runs as, for example 'us'. | string | No | — | — | — |
language Two-letter language code for the search, for example 'en'. | string | No | — | — | — |
location City-level origin for the search, to mimic a real user. | string | No | — | — | — |
num Maximum results to return on the page. Larger values are slower and drop some specialised result types. | number | No | — | 1–100 | — |
page Result page to fetch; skips the earlier pages of results. | number | No | — | ≥ 1 | — |
respond_with Set to 'no-content' to return only the result metadata, without reading each page. | string | No | — | one of "no-content" | — |
return_format Shape of the returned page content. | string | No | markdown | one of "markdown", "html", "text", "screenshot", "pageshot" | — |
engine Retrieval engine: 'browser' renders the page for best quality, 'direct' is fastest, 'cf-browser-rendering' is the experimental engine for JavaScript-heavy sites. | string | No | — | one of "browser", "direct", "cf-browser-rendering" | — |
timeout_seconds How long Jina waits for the page to load before giving up. | number | No | — | 1–300 | — |
links_summary Collect the page's links: 'true' for unique links, 'all' for every one. | string | No | — | one of "true", "all" | — |
images_summary Collect the page's images: 'true' for unique images, 'all' for every one. | string | No | — | one of "true", "all" | — |
remove_images Strip every image from the content (sends X-Retain-Images: none). | boolean | No | — | — | — |
with_favicons Include each result site's favicon. | boolean | No | — | — | — |
no_cache Bypass Jina's cache and fetch the page fresh. | boolean | No | — | — | — |
locale Browser locale used to render the page, for example 'en-GB'. | string | No | — | — | — |
Example configuration
Every value below is the contract’s own example or default — nothing invented.
{
"query": "jina embeddings pricing",
"return_format": "markdown"
}Output
One shape.
result
| Field | Type | Example |
|---|---|---|
results The search results, most relevant first. | array | — |
results[].title Result title. | string | — |
results[].url Result URL. | string | — |
results[].description Result snippet. | string | — |
results[].content The page, read into the return format. | string | — |
count How many results came back. | number | — |
query The query that was searched. | string | — |
{
"count": 0,
"query": "",
"results": [
{
"content": "",
"description": "",
"title": "",
"url": ""
}
]
}See Search the Web run in a workflow.
In a technical demo we add the Jina AI connection, drop the node on the canvas and let Workflow Health pre-flight it before the first real run.