Reproduction sequence
Extracts the visible sequence leading to a failure and separates observed behavior from interpretation.
VideoLens reviews screen recordings for visible state changes, spoken context, clicks, pauses, error messages, and failed outcomes—then produces findings tied to the moments that support them.
A screen recording often contains the best available evidence of a bug or confusing workflow, but someone still has to watch it, reconstruct the sequence, and write the ticket. VideoLens shortens that review loop.
Use Bug mode for reproduction steps and failure evidence, UX mode for friction and flow recommendations, or a custom prompt when you need a narrower investigation.
Extracts the visible sequence leading to a failure and separates observed behavior from interpretation.
OCR and frame descriptions preserve messages, labels, disabled states, and other interface evidence.
Produces a summary, severity hint, evidence citations, recommendations, and concrete follow-up tasks.
Every stage is explicit and cached so the source can be checked and the analysis can be reused.
It can produce a ticket-ready summary with reproduction steps, observed behavior, timestamped evidence, and suggested tasks. A reviewer should confirm severity, environment, and technical root cause.
It can read text from sampled frames. Increasing the sampling rate improves the chance of capturing brief messages but increases processing cost.
Yes. UX mode focuses on friction, pauses, repeated actions, confusing transitions, and recommendations rather than only failure states.
Try the hosted app, self-host the MIT-licensed core, or connect VideoLens to an MCP client.