Scrapeless
Developer ToolsScrapeless is a web scraping and unblocking platform. Agents run site-specific scrapers, unlock and render protected pages, crawl whole sites, and drive a scraping browser with sessions, profiles, and stored credentials.
What This Integration Enables
Scrapeless does four separable things, and the fourth one changes what kind of connector this is. It runs site-specific scrapers for Google, Amazon, and other named sources. It unblocks and renders individual pages. It crawls whole sites. And it drives a real Scraping Browser with sessions, saved profiles, uploaded extensions, and stored credentials, which means it can be signed in as you.
That last capability is a different category of thing from the first three. Reading a public page is a data question. Operating a browser that is authenticated as your company, on somebody else's platform, is a permissions question, and the honest version of the Orchestration as a Service argument says so out loud. Scrapeless is unusual in having designed for it: signals let an outside caller hand a named event with data into a browser session that is already waiting, and a live session URL lets a person watch what the automation is doing. That is a digital andon cord with the vendor's half already built. FlowRunner supplies the other half, which is the human on the end of it.
Without FlowRunner
With FlowRunner
Use Case Scenarios
-
Named-source data without a parser to maintain
Market and category work runs through the site-specific scrapers rather than through generic fetches, because someone else already keeps up with each site's markup. Scrape Google Search takes the query, locale, and domain, with one parameter selecting whether the results are images, local, video, or shopping. Scrape Amazon changes shape with its type field: product and seller take a URL, keyword search takes terms and a domain. Results land in Snowflake with the week's movement posted to Slack. When a scraper starts returning nothing, Get Actor Status answers the question a silent result raises, because site-specific scrapers break when the target site changes and this reports that rather than leaving a flow to infer it.
-
A documentation crawl that does not turn into a crawl of the internet
Crawl Website has a page limit that defaults to ten thousand, and every page is billed, so the limit is the spend control on every call rather than an optional tuning parameter. Three other settings decide whether a crawl of one section stays that: allowing external links lets it leave the site entirely, allowing backward links lets it climb above the starting path into the marketing site, and ignoring query parameters treats sorted and filtered variants of a page as one page, which on a faceted catalog is the difference between a hundred pages and ten thousand. When it finishes, the agent reads Get Crawl Errors as well as the status, because the blocked list names pages the crawler deliberately did not fetch and those look identical to pages that do not exist. The Markdown output feeds a knowledge base in Notion.
-
The supplier portal that has no API and never will
A profile carries cookies between sessions, so the login happens once and every later session that names the profile starts signed in. The credential behind it does not have to live on Scrapeless: the 1Password integration keeps the secret in your vault and resolves it when the session needs it, which is the better shape by a wide margin. The service-account token that makes that work reaches whatever the account can reach, so it belongs scoped to one vault rather than to everything. The session then does the work, and when it reaches something it should not decide alone, it waits.
Human-in-Loop Highlight
Scrapeless is the rare connector where the vendor built the pause and left the human out of it. Send Session Signal hands a named event with data into a browser session that is already running and already waiting, which the vendor's own documentation describes as how a flow passes a two-factor code or an approval into an automation rather than restarting it. List Session Signals answers what the automation is stuck on. Get Session Live URL shows a person exactly what the browser is looking at.
So the gate is not a message somebody skims. It is the release itself. The agent reaches the submit step inside the supplier portal, stops, and posts: "Session sess_4b21 is waiting on confirm_submit. It is signed in as the shared procurement account and is holding a purchase order for 1,400 units at 12.40, against quote Q-88213. Live view here. Send the signal to submit, or clear the signal and I will close the session." Whatever happens next is attributed to that account by the supplier, not to a scraper, and the portal has no undo. What made this safe was not that the agent navigated carefully. It was that the last click belonged to a person who could see the screen.
Agent Capabilities
45 actionsSite-Specific Scrapers
7- Run Scraper Runs any of Scrapeless's site-specific scrapers by naming the actor. The input means something different for every actor and the vendor documents each separately, which is why the common ones also have named operations below. With asynchronous mode on, the response is a task id rather than the data.
- Scrape Google Search Runs the Google Search scraper. Takes the query, interface language, country, and Google domain, with one parameter selecting the vertical: images, local, video, or shopping.
- Scrape Google Flights Runs the Google Flights scraper. The trip type field decides the shape of everything else: a round trip needs a return date, one way does not, and multi-city takes an itinerary array instead of departure and arrival fields.
- Scrape Google Scholar Runs the Google Scholar scraper for academic and citation results.
- Scrape Amazon Runs the Amazon scraper. The type inside the input selects what is scraped and changes the other fields with it: product and seller take a URL, while a keyword search takes terms, a page, and a domain.
- Scrape Shopee Runs the Shopee scraper. Its URL is frequently a Shopee internal API address rather than a page address, which is what the vendor's own examples use.
- Get Scraper Result Returns the result of a scraper run started asynchronously. The other half of every async scraper call, without which the task id is a receipt with nothing behind it.
Universal Scraping
2- Unlock Webpage Fetches a URL through Scrapeless's unblocking proxy and returns the raw response. The cheapest thing that works when a workflow only needs a page that blocks ordinary requests, with no browser and no session involved.
- Render Webpage Fetches a URL in a full browser so JavaScript runs, which is what a single-page application needs. The wait condition decides whether the content is there: the earlier events fire before a client-rendered app has drawn anything, so an empty result is usually this rather than a block.
Crawler
10- Scrape URL Scrapes a single page and returns it in the chosen formats. The one synchronous crawler operation, which makes it the right shape when a workflow needs the content in hand rather than a job id.
- Scrape URLs Starts a batch scrape of many pages. Asynchronous: the response is a job id, so the status and the errors are separate calls.
- Get Scrape Status Returns the status and result of a single scrape job.
- Get Batch Scrape Status Returns a batch scrape's progress and the pages it has finished. Reading completed against total matters more than reading the status, because a batch reports done having skipped pages.
- Get Batch Scrape Errors Returns the pages a batch failed on and why. A separate call from the status, so a workflow that only polls the status never learns that a third of its URLs never resolved.
- Cancel Batch Scrape Stops a running batch scrape. The stop-spending operation when a batch was handed the wrong URL list. Pages already scraped are billed and stay readable.
- Crawl Website Crawls a site from a starting URL, following links. The page limit defaults to ten thousand and every page is billed, which makes it the spend control on every call. Asynchronous, so the response is a job id.
- Get Crawl Status Returns a crawl's progress and the pages it has produced. Worth reading alongside the errors route rather than alone, because a crawl reports done having skipped most of a site.
- Get Crawl Errors Returns the pages a crawl failed on and the ones robots rules blocked. The blocked list is the useful part: it names pages the crawler deliberately did not fetch, which is indistinguishable from pages that do not exist.
- Cancel Crawl Stops a running crawl. The stop-spending operation, and it matters here because of that ten thousand page default.
Browser Sessions and Signals
8- Create Browser Session Starts a Scraping Browser session and returns its connection details. This is the route where the API token travels in the URL rather than in a header, so the connector logs the path only and masks the token out of error text.
- List Running Sessions Returns the browser sessions currently running on the account. The way a workflow finds sessions it forgot about, each of which is billing until its lifetime elapses.
- Get Session Live URL Returns a URL for watching a running session in a browser. The way to show a person what an automation is doing, and the fastest way to see why it has stopped.
- Send Session Signal Sends a named event with data into a running browser session. Scrapeless's rendezvous point: how a workflow hands a code, a value, or an approval to an automation already waiting inside the browser instead of tearing it down and starting again.
- Wait For Session Signal Blocks until a named event is signaled in a session. It holds the flow open, so a wait on something that never arrives stalls the execution until it times out.
- List Session Signals Returns the names of the signal events a session is currently waiting on. The answer to what an automation is stuck on, which the session itself does not report.
- Get Session Signal Stats Returns counts for a session's signal activity.
- Clear Session Signals Removes a session's pending signal events. Anything waiting on one of them stops waiting without receiving anything, so this is the abandon path rather than the release path.
Profiles and Extensions
8- Create Profile Creates a browser profile, which carries cookies and browser state between sessions. The way a login is reused: sign in once inside a session started with the profile, and later sessions that name it start signed in.
- List Profiles Returns the account's browser profiles.
- Get Profile Returns one browser profile.
- Update Profile Renames a browser profile. Its stored cookies are not editable through the API; they change by using the profile in a session.
- Delete Profile Removes a browser profile and the state it carried. Sessions that named it start signed out, which presents as a login that stopped working rather than as a missing profile.
- List Extensions Returns the browser extensions uploaded to the account.
- Get Extension Returns one extension's details.
- Delete Extension Removes an uploaded browser extension. Sessions that named it start without it.
Credentials
8- Get Credential Returns a team credential stored for an origin. This returns authentication data, so whatever a workflow does with the result is handling secrets.
- Create Credential Stores authentication data for an origin so a browser session can sign in with it. The secret is sent in the request body and stored on Scrapeless's side, which argues for a scoped, rotatable account rather than a person's own login.
- Update Credential Replaces the stored authentication data for an origin. Where a rotated password goes.
- Delete Credential Permanently removes stored authentication data. Sessions relying on it start failing to sign in.
- Get 1Password Secret Fetches one secret from your 1Password vault through Scrapeless's integration. The alternative to storing credentials on Scrapeless at all: the secret stays in the vault and is resolved when needed.
- Get 1Password Secrets Fetches several secrets in one call. Failures are per secret rather than per call, so each result needs reading individually rather than trusting the call's overall status.
- Set 1Password Token Stores a 1Password service-account token so Scrapeless can read your vault. The token carries whatever that service account can reach, which is the argument for scoping it to the single vault the automation needs.
- Revoke 1Password Token Removes the stored token. Sessions that resolve secrets from it start failing immediately, so this is a deliberate cutoff rather than a cleanup step.
Account
2- Get Account Returns the credit balance, plan, and usage. The operation to put in front of a crawl, since most of this API bills per request and the default crawl limit is what exhausts a plan. Credit figures come back as strings rather than numbers.
- Get Actor Status Reports whether the site-specific scrapers are currently working. Worth checking when a scraper starts returning nothing, because those break when the target site changes and this says so rather than leaving a workflow to guess.
Frequently Asked Questions
What can FlowRunner do with Scrapeless?
FlowRunner agents can run Run Scraper, Scrape Google Search, and Scrape Google Flights in Scrapeless, plus 42 more actions.
Does connecting Scrapeless to FlowRunner require OAuth?
No. Scrapeless connects to FlowRunner with an API key, no OAuth flow required.
Can Scrapeless trigger a FlowRunner workflow automatically?
Scrapeless doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.
Other Developer Tools integrations
Start building with Scrapeless
Free plan, no card required. Connect in minutes.