pick-mode-element-data
next
# A pick-mode highlight on a decorative element exports with an empty quote and no element data, which gives a reviewer nothing to point at. It happened twice in one review and the card had to be inferred from ordering alone.
# TODO: carry the element's own attributes, a stable selector path, and a curated set of computed CSS. CSS yes, JS no: listeners can be delegated, framework-attached, or minified past recognition, and there is no clean way to attribute them to an element. `getComputedStyle` returns roughly 340 properties, so a curated set only.
colour-schema-with-priority
next
# The five colours currently mean whatever you decide. Offer a schema where one axis encodes priority, and let the export group by colour before document order.
# TODO: the problem it solves is concrete. In one review the only real bug sat at item 4 of 7, between a formatting nit and a font-size note. Document order is not priority order, and a reader starting at the top starts in the wrong place.
count-notes-not-just-highlights
later
# The export header says `highlights: 12` and says nothing about how many carry a note.
# TODO: add `notes: N`. A highlight with a note is actionable; a bare highlight says only that somebody looked at something.
scope-markers, context-for-short-quotes
not a tool change
# Writing `all:` or `here:` at the front of a note, and adding surrounding context by hand when a quote is very short.
# Listed so nobody files them as missing features. These are things a reviewer does while writing, not things the tool should try to infer.
next = agreed for v1.0 · later = agreed, unscheduled · not a tool change = a way of working, kept here so it is not mistaken for a gap