Read one public URL through the server-side page reader and return clean Markdown plus structured images, colors, fonts, links, title, and description.
Where: Flowgen -> Add -> Web & Apps -> Read a Page
Before you start
- A public HTTP or HTTPS address reachable by the page-reading service.
Key ideas
- One address per run: Read a Page processes exactly one URL. For a list of addresses, connect an Items column through List iteration so Flowgen owns the fan-out, results, retries, and row identity.
- Server-side page access: The node uses the guarded server page reader rather than browser fetch. It can access cross-origin HTML that HTTP Request cannot, while applying SSRF protection and response/time caps.
- Readable and complete outputs: Page text emits Markdown for writers and research nodes. Everything found emits serialized JSON containing title, description, Markdown, images, colors, fonts, and links.
- Read a Page card: ScrapeUrlNode uses the authored shared-card path. Read one public URL into clean Markdown and structured page metadata.
- Properties sidebar: Card-owned scrape controls; common properties shell
- Opened surfaces: Markdown preview; structured page preview
Steps
- Type a URL on the card or connect one text value to url_in; a connected value wins.
- Turn on Let the page build itself first only when important content is rendered by JavaScript.
- Choose the image cap: 0, 6, 12, or 24.
- Run the node and inspect title, text length, image/color/font counts, and source link.
- Connect Page text to writing/research or Everything found to Transform, Items, Page Template, or another structured-data consumer.
Sockets
- URL / url_in: one text address.
- Page text / text_out: extracted Markdown.
- Everything found / json_out: serialized JSON with title, description, markdown, images, colors, fonts, and links.
Controls and limits
- Let the page build itself first: requests the rendered page; slower and more failure-prone, so it is opt-in.
- Maximum images: 0, 6, 12, or 24; default 12.
- One URL per run.
- The server caps response size and total read time.
Choosing a related node
- Use HTTP Request for an API or a known machine endpoint.
- Use Website Harvest for a broader site-level research workflow.
- Use Read a Page for one cross-origin page whose readable content and design facts are needed.
Read a Page controls and presets
- Add path: Flowgen → Add → Research & Insights → Collect → Read a Page.
- Card: ScrapeUrlNode.
- Properties: Card-owned scrape controls; common properties shell.
- Controls: URL; Let the page build itself first; Maximum images.
- Opened surfaces: Markdown preview; structured page preview.
Tips
- Start without rendered mode; enable it only if the returned Markdown is missing client-built content.
- Use Everything found when colors, fonts, images, or links matter—not only the prose.
- Keep URL, Page text, and data handles explicit in MCP-built flows.
Limitations and important notes
- Login-protected, blocked, or bot-resistant pages may fail.
- Rendered mode is not a full interactive browser session.
- The node does not crawl a whole domain.
- Returned external assets remain subject to their source permissions and availability.
Troubleshooting
The page text is nearly empty
Enable Let the page build itself first, then rerun. If it remains empty, the page may require login or block automated access.
Only some images were returned
Raise Maximum images. The node intentionally applies the selected cap.