# Website Screenshot & Change Monitor (PDF, Markdown) (`dgn-lab/website-screenshot-batch`) Actor Screenshot hundreds of URLs or a whole sitemap as PNG, JPEG or PDF, on desktop and mobile, and get alerted when a page changes: diff image in red, text added and removed, price and stock changes, Slack or Discord alerts. Markdown and JSON-LD for AI. Failed pages are free. - **URL**: https://apify.com/dgn-lab/website-screenshot-batch.md - **Developed by:** [Sinan Dogan](https://apify.com/dgn-lab) (community) - **Categories:** Developer tools, Automation, SEO tools - **Stats:** 2 total users, 1 monthly users, 100.0% runs succeeded, 0 bookmarks - **User rating**: No ratings yet ## Pricing from $6.00 / 1,000 page checkeds This Actor is paid per event. You are not charged for the Apify platform usage, but only a fixed price for specific events. Learn more: https://docs.apify.com/actors/running/actors-in-store.md#pay-per-event ## What's an Apify Actor? An Actor is a serverless cloud program that runs on the Apify platform. It has two run modes. In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours, and optionally produces a well-defined JSON output, datasets with results, or files in key-value store. In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server. Apify vocabulary and the platform model are defined once, in the agent quickstart at https://apify.com/agents.md. ## How to integrate an Actor? If asked about integration, you help developers integrate Actors into their projects. You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready. Do not guess an integration path. Every one of them is in the agent quickstart at https://apify.com/agents.md: the Apify MCP server, Agent Skills with the Apify CLI, the JavaScript and Python clients, the REST API, and the account-free path for an agent with no human to sign in. It also carries the rule on stating cost before the first paid run. For examples already wired to this Actor's own input schema, see the [API](#api) section below. Each client library has reference documentation the quickstart does not restate: [JavaScript/TypeScript](https://docs.apify.com/api/client/js/docs.md) (`npm install apify-client`) and [Python](https://docs.apify.com/api/client/python/docs.md) (`pip install apify-client`). # README ### What does Website Screenshot & Change Monitor do? **Screenshot hundreds of pages, and know the moment one changes.** **Website Screenshot & Change Monitor** captures **a list of URLs or a whole sitemap** as **PNG, JPEG or PDF** in a real [Playwright](https://playwright.dev/) browser, on desktop and mobile, then **compares every page with the previous run**: visual diff, text added and removed, price and stock. Three tools in one run: a **bulk website screenshot API**, a **website change monitor** with alerts, and a **web page to Markdown** extractor for AI. - 📸 **Clean screenshots in bulk:** desktop, Retina, iPhone, Pixel and iPad presets (several devices in one run), full page or a single element, PDF in A4, Letter, Legal or A3. Cookie banners, chat widgets and ads are hidden for you. - 🔍 **See exactly what changed:** share of changed pixels, a **diff image with the changes in red**, the lines of text added and removed, and ignore regions for clocks and carousels. - 🏷️ **Competitor price and stock tracking:** schema.org product price and availability compared from run to run, with the old and new values. - 🔔 **Alerts where you work:** Slack and Discord messages with the diff image, Telegram, or any webhook (n8n, Make, Zapier). - 🤖 **AI and RAG ready:** visible text as Markdown or plain text, metadata (canonical URL, language, Open Graph) and JSON-LD structured data. - 🔐 **Pages behind a login:** session cookies or HTTP headers, stored encrypted and sent only to the captured site. - 💲 **$6 per 1,000 pages, everything included.** Failed pages (blocked, not found, timed out) are never charged. **Who uses it:** | You are | You use it to | | ----------------------------------- | ------------------------------------------------------------------------------------------------------------ | | **E-commerce teams** | Watch competitor product pages and get alerted when a price drops or an item goes out of stock. | | **Agencies and QA teams** | Produce visual reports for clients and check every page of a site on desktop and mobile after each release. | | **Compliance, legal and marketing** | Archive dated captures of terms, pricing pages and ads, and be told when a competitor changes its messaging. | | **AI builders** | Turn sitemaps into Markdown for RAG pipelines and keep them fresh by re-crawling only what changed. | ### How to take website screenshots in bulk 1. Keep the two example **URLs**, or enter two public pages of your own. 2. Keep the custom 1280 × 800 viewport and PNG format for a quick first run. 3. Click **Start**, then inspect image links, dimensions and status in the dataset. 4. Open `screenshotUrl` to download each image. Download the dataset separately as JSON, HTML, CSV or Excel; these exports contain links and metadata, not an archive of image files. ```json { "startUrls": [ { "url": "https://docs.apify.com" }, { "url": "https://www.wikipedia.org" } ], "device": "custom", "viewportWidth": 1280, "viewportHeight": 800, "fullPage": false, "format": "png" } ``` ### Why use this screenshot API? - **One batch, explicit results:** successful files and failed URLs stay visible in one dataset. - **Mobile and desktop presets:** iPhone, Pixel, iPad and desktop/Retina settings include pixel density, touch and browser context. - **Several devices in one run:** select desktop and mobile presets together in **Several devices** to get every URL on each device. - **Sitemaps:** enter a `sitemap.xml` (sitemap indexes and `.xml.gz` files included) and filter its pages with include and exclude patterns. - **Content preparation:** optional scrolling, dark mode, hidden selectors and extra delay. - **Clean captures:** **Hide distractions** (on by default) removes live-chat widgets (Intercom, Drift, Crisp, HubSpot, Zendesk, Tawk.to, LiveChat, Tidio and more), hides ad slots while keeping their space, finishes animations and pauses videos. - **Dynamic sections:** wait for a visible CSS selector and optionally capture just that element as PNG/JPEG. - **Cookie-banner handling:** overlays of common consent platforms (OneTrust, Didomi, Cookiebot, Quantcast, Usercentrics, TrustArc, Sourcepoint, Axeptio and more) are hidden before the capture, and an accept button is clicked when found, including in consent frames and open shadow roots. Best effort, not a guarantee for every website. Disable it to preserve the initial consent screen. - **Long pages:** full-page captures taller than the image limit (about 60,000 pixels) are split into several images by default (`partUrls`), so nothing is lost. Choose **Cut at the maximum height** to keep one image, marked `truncated: true`. - **PDF settings:** A4, Letter, Legal or A3, portrait or landscape, with or without margins. - **Pages behind a login:** pass session **Cookies** or **HTTP headers** (stored encrypted, never logged, sent only to the captured site by default), and capture in another **Language** or **Time zone**. - **Download everything at once:** the **Output** tab links to a ZIP archive of all screenshots of the run. - **Change monitoring:** compare every page with the previous run, report `changed`, `changePercent`, `textChanged` and the changed lines, get a diff image, ignore dynamic regions, keep only changed pages, and get a chat or webhook notification. - **Text, metadata and structured data for AI:** visible page text as plain text or Markdown, description, canonical URL, language, first heading, Open Graph tags, and JSON-LD with a price and availability summary for products. - **Runs that finish:** blocked, timed-out or unreachable pages get a `failed` item with the reason; the run still succeeds, so schedules and integrations keep working. ### Capture pages from a sitemap Add a sitemap to **URLs**: any URL whose path looks like `/sitemap.xml`, `/sitemap_index.xml`, `/post-sitemap.xml`, or any `.xml.gz` file, is read instead of captured. Sitemap indexes are followed (up to 3 levels and 50 sitemap files), and gzip-compressed sitemaps are unpacked automatically. - **Sitemap: include URLs matching** keeps only pages whose URL matches a regular expression (case-insensitive), e.g. `/blog/` or `/products/[^/]+$`. - **Sitemap: exclude URLs matching** drops matching pages, e.g. `/tag/|/page/\d+`. - **Sitemap: max pages** (default 100, up to 1,000) caps the pages taken from sitemaps, in sitemap order. Ordinary start URLs are always captured on top, and a page that is both a start URL and in the sitemap is captured once. ```json { "startUrls": [{ "url": "https://docs.apify.com/sitemap.xml" }], "urlIncludePattern": "/actors/development/", "urlExcludePattern": "/input-schema", "maxSitemapUrls": 20, "format": "jpeg" } ``` Reading sitemaps is free; each page checked is charged as usual. The log and the `SUMMARY` record show how many sitemaps were read, how many pages they list and how many were selected. Sitemaps are downloaded directly, without the proxy. **HTTP headers** and **Cookies** follow the same scope as page captures (see [Capture pages behind a login](#capture-pages-behind-a-login)): headers go only to the host of the sitemap URL you entered, so a nested sitemap or a redirect on another host (a CDN, for example) never receives them unless **Send headers to every domain** is on; cookies go to the hosts they cover. A sitemap that cannot be read is logged as a warning and the run continues with the other URLs. Patterns are matched against the full URL, so check the paths the sitemap actually lists (they can differ from the address you see in the browser). ### Capture a dynamic section Set **Wait for element** to the selector that appears when the content is ready. **Capture element** saves the first matching element and overrides full-page image capture. Clear that selector for PDF output; PDFs always cover the page. ```json { "startUrls": [ { "url": "https://docs.apify.com" } ], "waitForSelector": "main", "captureSelector": "main", "format": "png", "selectorTimeoutSecs": 30 } ``` An absent element produces a failed item after **Element wait timeout**. The Actor disables CSS animations during image capture for more consistent files. ### Monitor pages for changes Turn on **Monitor changes** and run the Actor on a [schedule](https://docs.apify.com/platform/schedules), for example daily. The first run saves a baseline (`changed: null`). Each later run compares every page with its previous capture, for the same URL, device and capture settings, and reports: - `changed`: `true` when more than **Change threshold** percent of the pixels differ (any difference with the default 0), `false` otherwise. For PDF output only the text is compared, since PDF files change on every export. - `changePercent`: share of differing pixels, measured by the browser on the first image of the page. - `textChanged`: whether the visible text, of the page or of **Capture element**, changed. - `textChanges`: the lines of text `added` and `removed` since the previous run, in page order (up to 50 lines each, 300 characters per line; `truncated: true` when some were left out). The first 20,000 characters of each page's text are compared line by line. It is `null` when the text did not change, and on the first run after an update of the Actor, since earlier runs did not store the text. - `diffImageUrl`: a copy of the screenshot with the changed pixels highlighted in red, for changed PNG/JPEG pages. Very large captures (above 10 million pixels) get no diff image; `diffImageNote` explains why. - `previousCapturedAt` and `previousScreenshotUrl`: when and what the page looked like before (the link stays valid while the previous run's storage is retained). - With **Extract structured data**, `priceChanged`, `previousPrice`, `priceChangePercent`, `availabilityChanged` and `previousAvailability` (see [Monitor prices and stock](#monitor-prices-and-stock)); a price or stock change also counts as a change. **Only changed pages** keeps the dataset and storage small: unchanged pages get no item and no saved image. The previous capture of each page, with its text, is kept in a named key-value store (**Monitoring store name**, default `website-screenshot-monitor`) that persists across runs; use a different name for each independent watch list. Clocks, counters and carousels change on every run. **Hide distractions** (on by default) already removes chat widgets, hides ad slots and stops animations. For anything else, list the dynamic areas in **Ignore regions** (e.g. `#clock`, `.hero-carousel`): they are masked in a separate comparison image and their text is left out of `textChanged`, while the delivered screenshot still shows them. You can also capture a stable section with **Capture element**, hide elements entirely with **Hide elements**, or raise **Change threshold** (e.g. `0.5`). Ads loaded by video or outstream players that use no standard ad container may not be hidden; add their selector to **Ignore regions** or **Hide elements**. ```json { "startUrls": [{ "url": "https://example.com/pricing" }], "captureSelector": "main", "monitorChanges": true, "onlyChanges": true, "changeThresholdPercent": 0.5, "ignoreSelectors": ["#live-visitors", ".promo-carousel"], "stateStoreName": "competitor-pricing", "notificationWebhookUrl": "https://hooks.slack.com/services/..." } ``` #### Change notifications Set **Notification webhook URL** to receive one POST per run when at least one page changed. The URL is stored encrypted and never logged. A failing webhook is logged as a warning and never fails the run. | Destination | URL | Body | | ----------- | ------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------ | | Slack | Incoming webhook `https://hooks.slack.com/services/...` | Summary text, then the diff images (up to 10) | | Discord | Channel webhook `https://discord.com/api/webhooks/...` | Summary text, then the diff images as embeds (up to 10); mentions in page text never notify anyone | | Telegram | `https://api.telegram.org/bot/sendMessage?chat_id=` | Summary text with links to the screenshots and diff images | | Other | Any HTTPS endpoint (Make, Zapier, n8n, your API) | JSON: `event`, `changedCount`, `checkedCount`, `actorRunId`, `datasetId`, `pages[]` with `textChanges` | The summary lists each changed page with its share of changed pixels and its first text changes, above the links: ```text 1 of 3 monitored page(s) changed: - https://example.com/pricing (2.1% of pixels, text changed) − Pro plan $20/month + Pro plan $24/month https://api.apify.com/v2/key-value-stores/.../records/screenshot-....png changes: https://api.apify.com/v2/key-value-stores/.../records/diff-....png ``` Up to 3 removed and 3 added lines are shown per page; the dataset has the full list. Slack and Discord display diff images only when the key-value store is reachable by them (it is on the Apify platform); if Slack refuses an image, the message is resent without images. ### Read the change report Every run writes `REPORT` (Markdown) to the default key-value store; open it from the **Markdown report** output tab. It lists: 1. **Changed pages first**, largest change first: changed pixels (%), up to 5 removed and 5 added text lines per page, the price change (old → new) and stock change when price monitoring is on, and links to the screenshot and the diff image; 2. **failures**, with the reason for each page that failed or was skipped; 3. the **number of unchanged pages**. Text, prices and error messages come from the captured sites, so the report escapes them: a page cannot inject links, tables or HTML into it. Full details stay in the dataset (`textChanges`, `previousPrice`, `errorMessage`). ### Extract text, metadata and structured data **Extract page text** adds the visible text of the page, or of **Capture element**, to each result: plain text, or Markdown that keeps headings, lists, links and tables. Text longer than **Max text length** (20,000 characters by default) is cut and marked `textTruncated: true`. **Extract metadata** adds the meta description, canonical URL, language, first `h1` and Open Graph tags. All come from the same page load as the screenshot. **Extract structured data** reads the page's JSON-LD (`