Skip to content

Candor 6.11.0

Choose a tag to compare

@arunk3 arunk3 released this 18 Nov 22:33
· 3 commits to main since this release

Released 2026-11-18.

Figures and meters: StatCard's quiet empty, label style and accent figure, PortfolioCard's count in its status line and a footer, RunLog grouped by hour with a footer the app fills, UsageMeter's dashed mark when no limit is set, and a success tone the app can choose for spend within its limit.

With no CLI or @candor-ui/react change: only Pro items and the docs change.

Pro

  • StatCard and KpiRow take labelStyle="strong", the label at 11px, semibold, in the muted text colour. The
    default, "default", is 12px regular, as before.
  • A Kpi takes tone: "accent", which draws its figure in the accent text colour, for the one figure the person
    acts on. The colour is emphasis only, and forced colours drop it: say in the label or note that it is waiting on
    the person ("Waiting on you"). A figure not computed, or with nothing yet, is drawn as it always is. The standard
    metrics card drops a model's tone, as it drops its href: which figure waits on the person is the app's to say.
  • A Kpi takes empty, with value: null, for a metric with nothing to measure yet, such as an average with no
    runs. StatCard draws a small dash, hidden from screen readers, over your words, or "Nothing yet" with empty: true,
    and a screen reader hears those words in the figure's place. It is quiet, and never looks or reads like "Not
    computed", which stays the failure. missing wins over empty, since a failure must not read as calm, and a value
    wins over it; each logs a warning in development. A count of zero is value: 0. If you showed "None yet" through
    notComputed or valueText, move it to empty. The standard metrics card drops a model's empty, so a model's
    null still reads "Not computed": only the app knows there is nothing to measure.
  • PortfolioCard takes needsYouPlacement="status", which puts the count of things that need the person at the end
    of the status line, in words after the locale's separator ("On hold · 2 need you"), with no chip, for a card held on
    purpose. The count is still shown and read; zero follows allClear, and "Not checked" is always shown. The default,
    "chip", is as before.
  • PortfolioCard takes footer, one line of your words under everything else on the card ("Paused at stage 3, 14
    Nov"), drawn as plain text in an isolate. It is a string, not a slot, since the card is one link.
  • RunLog takes heading: a string in place of "Runs", or false for a page whose own heading names the log. The
    log keeps its name for screen readers either way, and with false its header row is left out when nothing else is in
    it (no toolbar, status filter, or Export once the runs are known). Blank words are no heading: the log reads "Runs".
  • RunLog takes groupBy="hour", which puts the runs under a heading for the hour they started, in
    LocaleProvider's time zone and date format, newest first, each heading with how many runs it holds ("6 Nov 2026,
    14:00 (3 runs)"). Before the page hydrates the hours are UTC's and say so, as every time in the log is, unless the
    provider sets both a locale and a time zone. An hour is the hour on the clock, in a zone a half or three quarters of
    an hour off UTC too; where the clocks go back and two hours read the same, each says its offset ("25 Oct 2026, 01:00
    GMT+1"). While RunLog's own status filter is on Failed or Running, a heading counts the runs shown ("1 shown"). A
    run whose start time can't be read goes under "Time not known", last unless groupOrder names it ("").
    groupBy also takes a function that returns each run's own key, with groupOrder and groupLabel, as
    DataTable's; a run it gives no key, or a group groupLabel gives no words, is under "No group", and an empty key
    logs a warning in development. With "hour", a key in groupOrder is the hour's start as a time, matched by its
    moment however it is written.
  • RunLog lists the runs newest first by the moment each started, whatever offset its time is written with, and a
    time it can't read last. It sorted startedAt as text, so mixed offsets sorted by their digits and a time that
    couldn't be read could come first. Runs whose times share one offset are in the same order as before.
  • RunLog takes footer, a line under the runs with your own figures: what they cover ("Today so far"), what they
    took and what they cost. RunLog never sums its rows: a number is written as each run's is, your text is shown as
    written, and null reads "Not measured". The line is hidden while RunLog's own status filter is on Failed or
    Running, since your figures cover every run.
  • UsageMeter's pill takes noLimitMark="dashed". Where the usage has no limit, it draws a thin dashed circle in the
    ring's place, an outline with no track and no fill, so it never reads as 0 % used, and keeps the ring's place, so the
    figure doesn't move when a limit is set or removed. It is hidden from screen readers and drawn in CanvasText in forced colours. With the prop,
    the pill's {noCap} words are the new pillNoLimitSet, "no limit set"; without it they stay pillNoCap's "no cap
    yet". Another variant ignores the prop, with a warning in development.
  • UsageMeter and UsageMeters take normalTone="success", which draws the normal level in the kit's success colour,
    held to 3:1 on the card as every mark is. Warning and danger keep their colours. The normal level is then said in
    words with a check, "Within the limit", in every variant (in the pill's words for screen readers, beside its check),
    so green is never the only sign. Where the levels fall stays yours: pass level from your service and every variant
    follows it. The default, "accent", is as before.
  • UsageMeter, UsageMeters and PortfolioCard's usage fill the normal level in the marks colour,
    --cd-primary-mark, as the kit's other marks have since 6.8.0, where they read --primary. Candor's themes draw as
    before; a brand that sets --cd-primary-mark now colours these too. usageLevelTones gives each level's fill as
    CSS, beside its token; UsageLevelTone.fill is optional, so a tone you build yourself needs none.

Some DOM your CSS may match

  • UsageMeter, UsageMeters and PortfolioCard's usage: at the normal level the used part's and the track's inline
    background or stroke read var(--cd-primary-mark,var(--primary)) where they read var(--primary).
  • StatCard: its label's <p> has data-slot="stat-card-label", with or without an icon, so you can match the label
    without its classes. Nothing else in its markup changes without the new props.
  • Only with the new props: StatCard's p[data-slot="stat-card-empty"] and p[data-slot="stat-card-empty-text"], and
    data-tone="accent" on [data-slot="stat-card-value"]; PortfolioCard's span[data-slot="portfolio-card-separator"],
    span[data-slot="portfolio-card-needs"][data-placement="status"] and p[data-slot="portfolio-card-footer"]; RunLog's
    tbody[data-part="group"], each with a th[scope="rowgroup"], and tfoot[data-part="footer"]. Without them the
    markup is as in 6.10.0.

New string keys

New words in new places, with English defaults. Plural keys take one and other; {label}, {count}, {hour} and
{offset} stay as they are.

  • statCard: nothingYet, emptyMark.
  • runLog: group, groupLoaded and groupShown (plural), hourOffset ({hour} and {offset}), noGroup,
    notMeasured, timeNotKnown.
  • usageMeters: withinLimit, pillNoLimitSet.