Let users mark important moments or screen areas while recording

[adilei]

Problem

Users may recognize an important moment, control, value, or region of the screen while recording, but currently have no low-friction way to tell the recorder “this matters.” Voice narration can provide context, but it may be imprecise about exactly when or where the relevant information appeared.

Proposed direction

Allow users to add an explicit marker during an active recording. A marker should be timestamped and preserved as meaningful evidence for analysis. Consider supporting visual markers as well, such as selecting, pointing to, or highlighting an important screen area without interrupting the recorded workflow.

Possible interaction models include:

• A configurable global shortcut that marks the current instant.
• A lightweight annotation overlay for selecting or highlighting a screen region.
• An optional short text or voice note attached to the marker.
• A visible, non-blocking confirmation that the marker was captured.

Requirements

• Do not steal focus or introduce blocking dialogs into the recording.
• Preserve the marker’s timestamp and any region/note through capture, bundling, sensitive-data handling, and analysis.
• Ensure markers provide actual evidence to the describer rather than creating empty events.
• Account for multiple displays, display scaling, and changing window positions for visual regions.
• Keep annotation UI and confirmation artifacts out of captured evidence where appropriate.
• Evaluate whether markers measurably improve analysis quality before choosing the final interaction.

Context

PR #39 explored reviving manual markers with a global shortcut and confirmation toast. That implementation is being closed while we consider how markers should integrate with the recording and analysis architecture more broadly.