Execution starts at the first unconnected block
UScraper finds a non-proxy block with no incoming connection and runs from there. Keep one obvious starting block per scraper.
Documentation
Learn UScraper from overview to every block
This documentation teaches the graph model first, then explains each block in its own reference page so users can build custom scraping and automation workflows without starting from a template.
Overview
UScraper runs a visual browser graph
A workflow is a sequence of connected blocks. It can drive the browser, wait for page state, branch on conditions, capture network data, call APIs, export results, and deliver them to a webhook or Discord.
Normal blocks follow the next connection
Navigation, interaction, wait, extraction, and action blocks follow their connected path. Webhook delivers run results at the end of a run.
Condition blocks branch
Element Exists, Element Visible, and Text Contains expose true and false connectors so one graph can handle alternate page states.
Proxy configuration is special
Set Proxy configures browser startup and is excluded from normal graph execution. It does not need to sit in the main chain.
Multiple inputs create internal loops
Navigate can hold multiple URLs and Type Text can hold multiple text values. Use Loop Continue at the end of the repeated work to advance to the next input.
Loop Continue is not End
Loop Continue advances the active multi-input loop and then follows its outgoing connection after the final input. End terminates the run.
Structured Export is row based
The row selector finds repeated items. Each column selector is evaluated inside each row, then written to CSV.
Network capture and API requests serve different jobs
Network/XHR Capture saves requests and responses from your browser session to Project Data. HTTP/API Request sends a configured request directly to an API and uses the response as data.
Share results with your team or another service
Webhook sends run results and data as JSON at the end of a run. Discord delivers run results and CSV, JSON, or Excel data to your team.
Workflow
A practical path for building a scraper
The safest way to build a custom scraper is to design against one page, verify selectors, and only then expand to URL lists or more complex branching.
Choose a target page
Open the page manually, confirm the data is visible, and inspect the DOM in Chrome DevTools.
Build the graph
Start with Navigate, connect interaction and wait blocks, then finish with extraction, export, Loop Continue for repeated inputs, or End.
Stabilize selectors
Use CSS selectors or XPath that survive refreshes. Prefer data attributes and repeated row containers over copied full paths.
Export and verify
Run one URL or one search term first, inspect rows, then use multiple inputs and append mode once the CSV shape is correct.
Recipes
Common graph patterns
Use these patterns as starting points. Swap selectors, waits, and export columns to match the site you are scraping.
Single page extraction
Navigate -> Wait for Element -> Extract Text -> EndBest for article fields, profile details, counters, or one-off values.
Listing to CSV
Navigate -> Wait for Element -> Structured Export -> EndBest for products, reviews, directories, tables, search results, and cards.
Search form
Navigate -> Type Text -> Click -> Wait for Element -> Structured Export -> EndBest when a user must enter a query before the result rows appear.
Multiple search terms
Navigate -> Type Text -> Click -> Wait -> Structured Export -> Reset Page -> Loop Continue -> EndBest when the same search form should run for many inputs. Reset the page or clear the input before Loop Continue.
Lazy loaded page
Navigate -> Wait for Element -> Scroll -> Sleep -> Structured Export -> EndBest when results appear only after the viewport moves down the page.
Optional popup
Element Exists -> true: Click close -> Export, false: ExportBest for cookie banners, modals, and intermittent empty-state panels.
Capture browser network data
Navigate -> Network/XHR Capture -> Click load results -> Wait for Element -> EndUse when a search, filter, or load-more action retrieves data through network requests. Inspect the captured requests and responses in Project Data.
Call an API directly
HTTP/API Request -> End (inspect response data)Configure the endpoint, HTTP method, headers, authentication, and body to match the API. Inspect the response before combining it with other workflow steps.
Deliver completed results to a webhook
Navigate -> Wait for Element -> Structured Export -> Webhook -> EndSend run results and data as JSON to a receiving service at the end of the run. Verify the delivered payload with a small test dataset.
Share a report in Discord
Navigate -> Wait for Element -> Structured Export -> Discord -> EndSend run results and CSV, JSON, or Excel data to your configured Discord destination so the team can review the report.
Block reference
Every supported block has its own page
Open a block page for its purpose, behavior, configuration fields, example flow, outputs, caveats, and practical usage notes.
Navigation
Navigation
Navigate
Opens one page or loops through a list of pages.
Read block docsNavigation
Go Back
Moves the browser back one step in its history.
Read block docsNavigation
Go Forward
Moves the browser forward one step in its history.
Read block docsNavigation
Refresh
Reloads the current page.
Read block docsInteraction
Interaction
Click
Clicks an element found by CSS selector or XPath.
Read block docsInteraction
Type Text
Clears and types one or more text values into an input field.
Read block docsInteraction
Select Option
Selects an option from a standard HTML dropdown.
Read block docsInteraction
Scroll
Scrolls the page or brings a target element into view.
Read block docsExtraction
Extraction
Extract Text
Extracts visible text or a single attribute from matching elements.
Read block docsExtraction
Extract HTML
Extracts inner or outer HTML from matching elements.
Read block docsExtraction
Extract Links
Collects href values from links.
Read block docsExtraction
Network/XHR Capture
Saves browser requests and responses to Project Data.
Read block docsWait & Timing
Wait & Timing
Wait for Element
Waits until an element exists or is visible.
Read block docsWait & Timing
Wait for Text
Waits until specific text appears on the page or inside an element.
Read block docsWait & Timing
Wait for Page Load
Waits for the base page document to be available.
Read block docsWait & Timing
Sleep
Pauses execution for a fixed number of seconds.
Read block docsConditions
Conditions
Element Exists
Branches based on whether an element exists in the DOM.
Read block docsConditions
Element Visible
Branches based on whether an element is present and displayed.
Read block docsConditions
Text Contains
Branches based on whether text appears on the page or inside an element.
Read block docsSetup & Actions
Setup & Actions
Set Proxy
Configures browser proxy settings before execution starts.
Read block docsSetup & Actions
Set Window Size
Resizes the browser window.
Read block docsSetup & Actions
Take Screenshot
Captures the current browser page as an image.
Read block docsSetup & Actions
Inject JavaScript
Runs custom JavaScript in the page context.
Read block docsSetup & Actions
End
Terminates the current scraper run.
Read block docsSetup & Actions
Loop Continue
Advances the active multi-input loop without ending the run.
Read block docsNotification
Notification
HTTP/API Request
Calls an API with headers, authentication, and a request body, and uses the response as data.
Read block docsNotification
Webhook
Sends run results and data as JSON to a webhook at the end of a run.
Read block docsNotification
Discord
Sends run results and CSV, JSON, or Excel data to Discord.
Read block docsQuality checklist
Before you trust a scraper
Run the same checks every time you build a new graph or update selectors for a changed website.
- Use one URL while designing and debugging the graph.
- Confirm selector match counts in DevTools before exporting.
- Use Wait for Element or Wait for Text instead of long fixed sleeps when possible.
- Place Loop Continue after the reset step that prepares the next multi-input iteration.
- Use append mode only after confirming headers and columns are correct.
- Respect site terms, robots rules, rate limits, privacy, and data ownership requirements.
Windows 11
macOS 12+
Linux