Can Valmera record a website from its URL?

Yes, for accessible public pages. A pan holds and scrolls the page; a scripted walkthrough can click, hover, type, scroll, wait and navigate. The result is a video asset, which you then place and edit. A URL alone does not guarantee that every requested interaction will succeed.

Is the example on this page an actual recording?

Yes. It was captured through Valmera MCP on September 9, 2026. The published successful file is 10.833333 seconds, 1920 × 1080, H.264 at 30 frames per second, with no audio stream. Two earlier attempts skipped the requested click. This is a browser capture, not a finished Studio export or a comparative benchmark.

Can it record my signed-in dashboard or native app?

The browser recorder does not inherit your signed-in tabs and will not type into password or payment fields. Record an authorized private dashboard or native app yourself, preferably with synthetic demo data, then upload the footage. A public sandbox must actually be accessible without your private session.

Why can it miss a link that is visibly on the page?

A label or selector can match multiple elements, including a hidden mobile navigation link. In the published example, a label and a broad selector both failed; a selector scoped to the visible desktop sidebar succeeded. Inspect the current page structure and the failed-step report before correcting the target.

Does it always draw a visible click ring?

The successful example has a visible ring in inspected frames at 3.6 and 4.0 seconds. That observation replaces this page's older claim that click flashes never survived capture. It is evidence for one recording, not a guarantee for every site, browser state or capture.

Does recording automatically add zooms and click sounds?

No. Recording creates the asset and event information. Placement and editing are later operations. Positioned click events can support aimed zooms; click times alone do not identify a target. Sounds are separate additions. Inspect what the editing operation actually returned and rendered.

Does the browser capture include page audio or speech captions?

The browser capture is silent. The published example has no audio stream, despite a generic handoff label mentioning AAC. Add narration, music or timed text separately. A silent capture has no spoken transcript, and separately added narration needs its own caption and synchronization review.

How do I get the final video?

After placing and editing the capture, review the intended edit version and create a new final MP4 in Studio. The current external MCP contract keeps final-export creation in Studio. Account creation and uploads are free; paid access and credits are required for AI editing. There is no direct social publishing or interactive HTML demo export.

Record a website demo from a URL

Valmera can record an accessible public web page as video. Describe a short walkthrough, inspect the capture and its reported events, then place it in your edit. The example below shows what worked, what failed and what the delivered file contains.

Watch the actual capture

Recorded September 9, 2026: the pointer moves toward the desktop Captions link, the browser opens that documentation page, then scrolls down. This is a raw browser capture, with no added narration, zoom editing or Studio final export. It is silent.

Open the dedicated watch page · Download the successful capture · Download the first, incomplete attempt · Requests and evidence JSON

The successful file measures 10.833333 seconds, 1920 × 1080, 30 fps and 325 frames, encoded as H.264. It has no audio stream. Its 875,287 bytes were preserved when publishing this copy; the evidence JSON contains its SHA-256 hash.

Two skipped clicks before a successful walkthrough

The first request used the label “Captions.” The next used a broad link selector. Both produced a video while reporting that the click was skipped. A third request selected the desktop sidebar link specifically and reported one click and one scroll, with no skipped step.

All three capture attempts on the same documentation site
Attempt	Click target	Result reported by the recorder
1	Visible-label request: Captions	13.0-second capture; zero clicks; two scrolls; requested click skipped.
2	a[href='/docs/captions']	11.7-second capture; zero clicks; one scroll; requested click skipped.
3	.docs-sidebar a[href='/docs/captions']	10.8 seconds reported; one click and one scroll; no skipped step reported. File duration measured separately as 10.833333 seconds.

The current page has a hidden mobile link before the visible desktop link. Local recorder source selects the first match, which is consistent with these failures; this does not independently verify the deployed implementation. The practical correction demonstrated here was a selector scoped to the visible sidebar. Page structure changes, so this selector is an example for this page, not a universal recipe.

Exact successful capture request for an MCP client

Call record_website_demo in your own project. Replace project_id with the identifier returned by your project-creation or project-opening call.

{
  "project_id": "YOUR_PROJECT_ID",
  "url": "https://valmera.io/docs/getting-started",
  "orientation": "landscape",
  "steps": [
    {
      "do": "wait",
      "seconds": 1.5
    },
    {
      "do": "click",
      "selector": ".docs-sidebar a[href='/docs/captions']",
      "seconds": 2
    },
    {
      "do": "scroll",
      "by": 350,
      "seconds": 2
    },
    {
      "do": "wait",
      "seconds": 1.5
    }
  ]
}

How to record and edit a public website walkthrough

1
Choose a reachable public page
Use a page you have permission to record. The recorder opens a fresh visitor session, without your existing login. Keep the demonstration to a few read-only actions and avoid creating real accounts, invitations or purchases.
2
Describe the walkthrough
Name the starting URL, orientation, actions and useful pauses. Use visible link or button labels; when a label is ambiguous, provide a selector for the intended visible element. A pan is a separate capture mode for holding and scrolling a page without clicking.
3
Inspect the capture and reported events
Wait for the capture job. Read any skipped-step or page-loading report, then play the actual file. Confirm the intended navigation, pointer, scrolling and important text. A completed file can contain an incomplete walkthrough.
4
Place, edit and deliver
The capture becomes a project asset; recording alone does not place it on the timeline. Place it, review any requested zooms, add narration or text if wanted, and create the edited final MP4 in Studio. Check the downloaded final before sharing.

A scrolling pan uses record_website, whose exposed contract allows 4–30 seconds and an optional stationary hold. A scripted walkthrough uses record_website_demo. Choose the shorter sequence that communicates the task; loading and execution time affect its length. Check current tool schemas and returned limits rather than promising a fixed duration.

A useful initial brief is: “Record this public documentation page, open the named guide, scroll to the relevant example and hold long enough to read it. Report any step you could not complete.” Requests are proposals until the capture has been checked.

Read the event track on the capture's clock

Reported events and independent observations in the successful capture
Evidence	Time or position	What it establishes
Reported click	3.57 seconds; normalized position (0.06, 0.20)	The recorder reported a press at the sidebar target in this capture.
Inspected frames	3.5, 3.6 and 4.0 seconds	Pointer near the link, a small ring, then the Captions page with a larger ring. Sampled visual observations, not a universal timing guarantee.
Reported scroll and later frame	Scroll at 5.72 seconds; frame inspected at 10.0 seconds	The recorder reported a scroll; the later frame shows lower content on the Captions page.

These seconds refer to the captured clip. After placing it in a larger edit, output time also depends on its position, trims and speed. Click times alone do not give a button location; positioned events can support aimed movement. Inspect any requested zoom after placement.

The older version of this page said click flashes did not survive capture. The inspected frames from this run show a visible ring, so that absolute statement has been removed. The file probe also found no audio stream, despite a generic handoff description mentioning “H.264 + AAC.” File contents take precedence over a generic label.

What this workflow does and does not cover

Public visitor access: no inherited login, desktop-app recording or native mobile capture. Password and payment-field typing are refused. Record private sessions yourself with demo data.
Page-dependent results: missing targets, consent screens, bot checks, redirects or loading problems can change the footage. A successful capture job is not proof that the intended page and actions were recorded.
Separate placement and editing: a saved asset is not automatically on the timeline. Click sounds and aimed zooms are later work; this example did not test those operations.
Source scope matters: a main-source cursor or region operation does not establish that every inserted capture is processed. Fixed blur regions do not track moving text or censor inserted clips. Prefer clean demo data.
Silent footage: add narration, music or timed text separately. A silent capture has no speech transcript; separately added narration needs synchronization, caption and mix review.
Output review: a portrait browser viewport is not a physical phone. Crops, rounded frames and overlays can make UI labels unreadable. Review each destination size.

For the complete product-video workflow, use the SaaS demo guide. The current external MCP contract keeps new final-export creation in Studio. The edited final is newly encoded; Valmera does not export interactive HTML demos or publish directly to social platforms. Account creation and uploads are free; AI editing requires paid access and credits. See current plans.

Method and evidence limits

Valmera publishes this guide and operates the recorder. We made three direct MCP capture calls against one public documentation site. The first and third files were downloaded for inspection; the second attempt's duration and outcome are tool-reported. All 325 frames of the successful file decoded, and four selected moments were visually inspected. These checks do not amount to continuous human review of every frame or an independent event-timing benchmark.

No private dashboard, phone, cursor-enhancement pass, zoom edit, voiceover, comparative speed test or new Studio final export was tested here. The successful capture file is published unchanged. The requests and failed attempts remain visible so the example does not imply first-attempt success.

Frequently Asked Questions

Record a public walkthrough, then review it

Start with a short public visitor flow. Inspect the capture, place it in your edit and create the final in Studio. Paid access and credits are required for AI editing.

Create an account

See pricing →