UNGUESSUNGUESSAccessibility report assistant

How to use this tool

About a minute per bug once you have the evidence in front of you.

  1. Describe what you observed

    Which element, on which page or screen, and with which assistive technology. If a screen reader is the point, write what it announced. This text is what drives the criterion mapping, so be concrete: two specific sentences beat five vague words. A very short description makes the tool skip the shortlist and weigh all 86 criteria instead, which is slower and no more accurate.

  2. Attach the evidence

    Up to 5 screenshots, plus video. Video is processed in your browser: it extracts up to 7 key frames, drops the near-identical ones, and sends only those. The video file itself never leaves your machine.

  3. Pick the report language

    Italian or English. It changes the generated report, not this interface.

  4. Run the analysis

    One call to the model, images included. The figure above the button is an estimate; the real cost reported by OpenRouter appears with the result.

  5. Read the report, then edit it

    Title, steps, expected and actual result, remediation and severity are all editable. The model proposes, you stay the author — nothing is exported that you have not read.

  6. Add it to the queue

    The queue lives in your browser and survives closing the tab, so you can work through a session and export once at the end.

  7. Export and paste

    The xlsx has one column per bugform field and nothing else: no empty platform columns, row heights already fitted to the text. Markdown and JSON are there for tickets and for scripts.

Accessibility report assistant

Describe what you found, attach screenshots or video, and get a structured bug report mapped to the right WCAG 2.2 success criterion, in the UNGUESS format.

New finding

What you observed, on which element, and with which tool. The more concrete it is, the more precise the criterion mapping.

Attachments

Drop screenshots or video here, or choose files

PNG, JPG or WebP images and MP4 or WebM video. Images are resized to 1568px and key frames are extracted from video: the original video never leaves your computer.

Reading the list from OpenRouter…

Estimate~2,300 input tokens

Bug queue

0 bugs queued

Nothing queued yet. Reports you add stay here even if you close the browser.