Visual feedback glossary

Feedback delivered with screens, not just words.

Grabbing a still or sequence from a screen, tab, or window.

A single image of a page taller than the viewport.

A capture of every pixel a page would render if scrolled.

A recorded video walkthrough of an on-screen workflow.

Marks added on top of a screenshot to direct attention.

Trimming a capture down to the rectangle that matters.

Review that doesn't require everyone in a meeting.

A shareable link that is the feedback artifact itself.

The exact page address where a finding was observed.

The artifact that turns an observation into an actionable ticket.

The minimal sequence that reproduces a defect on demand.

The checklist that decides if a story is done.

Behavior that used to work and now doesn't.

Input or state at the boundary of normal behavior.

How bad it is vs how soon to fix it.

Sorting incoming defects into what gets worked on next.

A quick check that the build isn't fundamentally broken.

The default happy-case route through a feature.

The shared bar a work item must clear before it ships.

Structured critique of a design against goals and standards.

Peer feedback on a design in progress, not a sign-off.

Designers verifying that dev built what design specified.

Inspecting a UI against a fixed list of usability rules.

Watching a real person attempt a task on the product.

Observations about how a workflow actually feels.

Stepping through a task as a first-time user would.

A capture marked up with arrows, text, or boxes.

An LLM-powered tool that writes and edits code on request.

The body of input an agent uses to ground its next action.

The text supplied alongside an instruction to an LLM.

A format an LLM can parse and act on without preprocessing.

Saving content as plain markdown for downstream use.

Putting a markdown representation on the clipboard in one click.

Retrieved snippets fed into an LLM at prompt time.

Passing work to the next actor in a defined, parseable shape.

Whatever an agent reads to start a task.

The maximum tokens an LLM can attend to in one request.

The visible browser area, sized by the device.

A width at which a layout switches to a different design.

A color scheme with light text on dark backgrounds.

The browser panel that prints JavaScript errors and logs.

The browser panel that lists every request a page made.

Lossless screenshots vs smaller, lossy photos.

The unambiguous specification of markdown syntax.