The 22 skills · Craft skill · Core Pack and Everything

AGENT SKILL

glance-test: what it makes your agent do, and what it refuses.

glance-test governs the last gap in reporting: the distance between a correct page and a correctly read page. An executive gives a dashboard about twenty seconds, and a figure that is perfectly computed can still be misread in eight of them because the layout invited it.

The job glance-test does

The skill designs the landing on purpose. One number the page exists to deliver, its comparison riding with it rather than a row away, then the two things that actually moved. Above the content sits a strip that leads with what was proved, in words the reader could repeat. Any figure contained inside another renders subordinate and says so, because two equal tiles invite a reader to add them. A sample size travels with its own figure rather than sitting in the page footer, and the page states what was checked and what was not in the reader's language. Before shipping, the skill runs a twenty second dry run against its own page and answers five questions out loud.

What glance-test refuses

A skill is worth what it stops as much as what it produces. These are the refusals written into glance-test, taken from the rules and limits in its own file.

• ✓A page where everything is emphasised. If nothing leads, the eye picks by accident and the page has chosen for the reader at random.
• ✓A bare value. A number with no comparison answers nothing, so the delta and its tone sit with the figure.
• ✓A component rendered as a peer. A figure inside another is drawn subordinate and labelled part of, so the page never invites a double count.
• ✓A page level sample size. An n beside one metric and computed on another is a wrong number with a right value.

8 lines from the glance-test skill file, verbatim

From skills/glance-test/SKILL.md, unedited. Kymira skills are plain text you can read in full before you install them, which is the point of shipping a method rather than a binary.

1. What is the **one** number? (If you can't tell in two seconds, the
   hierarchy failed.)
2. Is every headline number sitting next to its comparison?
3. Could a hurried reader **add** two tiles that shouldn't be added?
4. Does every sampled or estimated figure carry its own `n` / caveat,
   inline?
5. Does the honesty strip lead with what was proved, in words the reader
   could repeat?

The failure modes glance-test addresses

These entries in the catalogue of 51 observed failure modes name this skill as what addresses them. Every one was seen in a real build or in our own audits, and none was invented.

The tier glance-test ships in

Both paid tiers are a one-time purchase with lifetime updates, and all sales are final: the files are plain text you keep the moment they arrive, so judge the quality on the free skill and the published doctrine first. The pricing section states exactly what each tier contains.

Installing glance-test

A skill is a folder with a SKILL.md file in it. Installing one is copying that folder where your agent looks for skills, with no build step and no dependencies.

mkdir -p ~/.claude/skills
cp -R Kymira/skills/glance-test ~/.claude/skills/

Or open the unzipped pack in Claude Code and say install Kymira, which installs every skill in your tier, or double-click the installer for your system. Cursor reads the same skill as a rule file: copy glance-test.mdc into .cursor/rules/. Skills are read at startup, so start a new session and ask your agent which skills it has available.

Not for

Last updated 2026-08-09 · Dylan, founder · one of 22 packaged skills, each one a method written down, not a product tour.