Title99/100
Scrape a URL
The title clearly communicates the tool’s action and subject. Keep it concise.
Description98/100
Scrape one known URL and return any combination of HTML, Markdown, a screenshot, page images, original bytes, CSS-selected fields, relevant highlights, schema-shaped JSON, or product data. Supports browser actions, custom target headers, PDF parsing, deadlines, and zero data retention. Browser clicks can change the target site's state. Prefer responseMode="compact" with formats.highlights and highlightsParams.query for focused research; use full for complete content or metadata. Use web-search when the URL is unknown, web-map for URLs only, web-crawl for linked pages, or submit-batch for large asynchronous jobs.
The description explains the capability, when to use it, and what to expect.
Inputs90/100
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
url
Public HTTP or HTTPS URL to scrape.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
formats › html
Rendered HTML.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
formats › markdown
Page content as Markdown.
Clear enough to supply this argument.
formats › screenshot
A screenshot of the page.
Clear enough to supply this argument.
formats › images
Images found on the page.
Clear enough to supply this argument.
formats › bytes
The original HTTP response body.
Clear enough to supply this argument.
formats › parse
Fields extracted with `parseParams.rules`, returned as `parsed`.
Clear enough to supply this argument.
formats › highlights
Markdown excerpts relevant to `highlightsParams.query`.
Clear enough to supply this argument.
formats › json
An object matching `jsonParams.schema`, extracted from the page.
Clear enough to supply this argument.
formats › product
Product details such as name, price, and availability.
Clear enough to supply this argument.
markdownParams › includeLinks
Keep link URLs in the Markdown. Set false to return link text without URLs.
Clear enough to supply this argument.
markdownParams › includeImages
Include images in the Markdown using image syntax with URLs and alt text.
Clear enough to supply this argument.
markdownParams › inlineImages
How base64 images appear: `placeholder` (default) or `preserve`. Requires `includeImages`.
Clear enough to supply this argument.
screenshotParams › area › anyOf › 1 › selector
CSS selector matching exactly one visible element.
Clear enough to supply this argument.
screenshotParams › area › anyOf › 2 › x
Left edge of the capture, in pixels from the document origin.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
screenshotParams › area › anyOf › 2 › y
Top edge of the capture, in pixels from the document origin.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
screenshotParams › area › anyOf › 2 › width
Width of the capture in pixels.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
screenshotParams › area › anyOf › 2 › height
Height of the capture in pixels.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
screenshotParams › format
Image format for the screenshot.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
imageParams › enrich
Extra data per image: `dimensions`, `classification`, or a hosted `file` URL.
Clear enough to supply this argument.
imageParams › dedupe
Set `visual` to drop visual duplicates, keeping the largest copy.
Clear enough to supply this argument.
parseParams › rules › additionalProperties › anyOf › 1 › selector
CSS selector to match within the current page or parent rule.
Clear enough to supply this argument.
parseParams › rules › additionalProperties › anyOf › 1 › type
Return the first match with `item` or all matches with `list`.
Clear enough to supply this argument.
parseParams › rules › additionalProperties › anyOf › 1 › output
Return text, HTML, an attribute such as `@href`, or nested field rules.
Clear enough to supply this argument.
highlightsParams › query
The question or topic to find passages for.
Clear enough to supply this argument.
highlightsParams › maxCharacters
Maximum combined length of returned passages.
Clear enough to supply this argument.
jsonParams › schema
JSON Schema (not an example object) for a top-level object, up to 50 KB. Use optional or nullable fields for missing facts.
Clear enough to supply this argument.
jsonParams › instructions
Extra guidance, such as which facts to prefer or how to read a field.
Clear enough to supply this argument.
productParams › dedupeImages
Drop visually duplicate product images, keeping the largest copy.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › country
Proxy country as a two-letter code, such as `US`. Case-insensitive.
Clear enough to supply this argument.
sharedParams › headers
HTTP headers to send to the target site. Requests with headers skip the cache.
Clear enough to supply this argument.
sharedParams › viewport › width
Browser viewport width in pixels.
Clear enough to supply this argument.
sharedParams › viewport › height
Browser viewport height in pixels.
Clear enough to supply this argument.
sharedParams › theme
Emulate a light or dark color scheme.
Clear enough to supply this argument.
sharedParams › dismissCookies
Accept cookie banners before actions and capture.
Clear enough to supply this argument.
sharedParams › dismissPopups
Close other popups before actions and capture.
Clear enough to supply this argument.
sharedParams › actions › items › type
Use `perform` for a plain-language browser instruction.
Clear enough to supply this argument.
sharedParams › actions › items › action
One browser instruction, such as clicking a button or entering text.
Clear enough to supply this argument.
sharedParams › actions › items › direction
Direction to scroll.
Clear enough to supply this argument.
sharedParams › actions › items › amount
Distance per scroll: pixels, one `viewport`, or `max` to reach the end.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › actions › items › selector
Scroll this container. Omit to scroll the page.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › actions › items › maxScrolls
Maximum number of scroll steps for this action.
Clear enough to supply this argument.
sharedParams › actions › items › milliseconds
Time to pause in milliseconds before the next action.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › waitFor
Milliseconds, or a CSS selector to wait for, after actions.
Clear enough to supply this argument.
sharedParams › settleAnimations
Wait for CSS animations to finish before capture.
Clear enough to supply this argument.
sharedParams › mainContentOnly
Keep only the main content. Doesn't affect `screenshot`, `bytes`, or `product`.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › includeSelectors
Keep only elements matching these CSS selectors.
Clear enough to supply this argument.
sharedParams › excludeSelectors
Remove elements matching these CSS selectors. Overrides `includeSelectors`.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › includeFrames
Include iframe content in HTML and text outputs. Screenshots always show visible frames.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › parsers › pdf › startPage
First page to parse, starting at 1.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › parsers › pdf › endPage
Last page to parse. Must be at least `startPage`.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
sharedParams › parsers › pdf › ocr
Set `auto` to read scanned pages with OCR.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
maxAgeMs
Maximum age of a cached output, in milliseconds. `0` fetches fresh. Maximum: 1 year (31536000000 ms).
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
zdr
`enabled` turns on zero data retention. Your organization must have ZDR enabled.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
timeoutOpts › milliseconds
Deadline in milliseconds.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
timeoutOpts › behavior
"fail" returns 408 at the deadline. "return-partial" returns available results; inspect the response’s partial flag. "return-partial" requires at least 5000 ms.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
tags
Labels for tracking request usage. Not retained when zdr is enabled.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
responseMode
Prefer compact for research: one JSON text response, no unused formats or verbose page metadata; HTML/Markdown/Highlights capped at 1,200 characters per field per search result or 4,000 per scrape or crawl field. Clipping is marked. JSON, parsed data and images are retained. Full (default) keeps the existing response shape and server limit. Use Highlights for query-relevant excerpts instead of clipped page prefixes.
Remove validation rules already expressed in this field’s schema; describe its meaning and usage instead.
Annotations75/100
Declare idempotentHint based on the tool’s actual behavior.
{
"readOnlyHint": false,
"openWorldHint": true,
"destructiveHint": true
}Output schema0/100
No output schema is declared. Add one for structured results so clients can interpret returned fields.
Not declared