Ask About a Screenshot of the Current Page

Capture & Analyze takes a screenshot of the current tab and attaches it to your next question in the WTFAA side panel. It is useful when the answer depends on what the page looks like rather than only on its text: a layout, a chart, an error message, or an image.

The button sits in the input area next to the message box, not on the role toolbar, so it is available in every role: Reader, Designer, and Developer. WTFAA captures a screenshot only when you click the button.

Before You Start

• AI access: You need a working AI service, either Platform (with a Pro subscription) or Use My Own Keys. See Getting started with WTFAA if you have not set this up.
• A model that can read images: When you use your own keys, the Google Gemini, OpenAI, and Anthropic Claude models in the model list accept images. The DeepSeek and Groq models in the list work with text only. See "When the Model Cannot Read Images" below.

Capture Part of the Page

1. Open the page in a normal Chrome tab and open the WTFAA side panel.
2. Scroll the page so the part you want to ask about is visible in the tab. The screenshot covers what is currently shown in the tab's page area. It does not include the parts of the page you would have to scroll to, and you cannot draw a selection around a smaller region.
3. Click Capture & Analyze (the camera icon next to the message box).
4. A thumbnail of the screenshot appears above the message box. Click it to view it larger, or click the remove button on the thumbnail to discard it. Capturing again replaces the previous screenshot.

The screenshot is scaled down so that its longest side is at most 1,280 pixels before it is attached.

Ask Your Question

1. Type your question in the message box, for example "What does the error message in this screenshot mean?" A screenshot cannot be sent without a message.
2. Press Enter or click send. Because a screenshot is attached, the Review before sending dialog always opens first.
3. Check the dialog. It shows your message, whether the screenshot will be sent, the model, the mode (Platform or BYOK), and, if page context is attached, the page URL, title, and text length.
4. Adjust the options if needed:
   • Send screenshot: Clear this box to send the question without the image.
   • Send page or form context: Clear this box to leave out the page or form text that would otherwise go with the question.
5. Click Send to AI.

If you click Cancel, nothing is sent and both your message and the screenshot stay in the panel. After a message is sent, the screenshot is removed from the input area, even if you chose not to send it. Capture again for the next question.

Quick actions on the role toolbar do not attach the screenshot. To ask about a screenshot, type your own question.

When the Model Cannot Read Images

If you use your own keys and the selected model cannot read images, WTFAA does not send the screenshot. It sends your message as text only and shows this warning in the chat:

⚠️ Current model does not support image analysis. Image ignored, text only sent.

The answer is then based only on your message and any page context. To include the image, choose a model from Google Gemini, OpenAI, or Anthropic Claude in Settings, then capture and send again.

Pages Where Capture Is Not Allowed

Chrome does not let extensions capture some pages, such as browser system pages and the Chrome Web Store. On those pages, clicking Capture & Analyze shows "Failed to capture screen". Switch to a regular web page and try again.

Privacy Notes

A screenshot contains everything visible in the tab at that moment, including names, email addresses, account details, or other personal information shown on the page. The local redaction in the send preview applies to page and form text; it does not change the image. Check the thumbnail before you send, and clear Send screenshot if the image shows something you do not want to share.

For more on the send preview, see Privacy and the send preview. The full privacy policy is at datafrog.io/privacy.