Skip to content
SkillModel-led2,292 tokens in use

interface-review

Reviews a branch, pull request or uncommitted change for the interface problems it introduced or made worse, with every finding classified.

12,499 stars on the repository MIT Licensejakubkrehel/skills · skills/interface-review
npx skills add jakubkrehel/skills
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing interface-review costs

38tokens idle
2,292tokens in use
5,080tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

shelldetectednetworkdetectedwrites filesdetectedgitdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Good
  • Frontmatter8/15
  • Structure19/25
  • Examples6/15
  • Output15/15
  • Size15/15
  • Resources10/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 75
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 75

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity96
  • Maintenance52
  • Community48
  • Maturity24
  • Momentum40

interface-review checks a change, where most UI reviews check a screen. It resolves which commits and files are in scope, expands the changed files to the surfaces that render them, reads the removed lines of the diff for lost accessibility and typography signals, and labels each finding as Introduced, Regression or Pre-existing.

The review is read-only. It never checks out branches or stashes work, and it hands the actual domain judgement and the final verdict to the better-interface skill. You invoke it yourself by command.

What You Get

  • A scope block naming the target, base and head refs, commit counts, files in scope, exclusions and surfaces expanded
  • A findings table with severity, domain, status, location, before, after and reason
  • Regressions found by scanning removed lines for aria attributes, focus styles, reduced-motion queries, logical properties and similar signals
  • Up to three pre-existing findings, stated as outside the change’s responsibility
  • Checks that the change delivers its stated intent, such as missing hover, focus, disabled or loading states and untranslated strings

Common Use Cases

  • Reviewing a pull request for accessibility or layout regressions before merge
  • Checking a feature branch that adds a new component for missing states
  • Checking uncommitted work for removed aria-label or focus outlines
  • Reviewing a design-token change for the consumers it affects

What Makes It Unique

It reviews a diff and its blast radius. Findings are statused by cause, so a problem a changed token creates in an untouched component counts as Introduced. It reads the removed side of each hunk against a table of signals and the equivalent replacements that clear them, handles awkward states such as shallow clones, detached HEAD and mid-rebase trees, and refuses to invent a change when none exists.

Under The Hood

The skill is written instructions plus two reference files, one on removed signals and one on resolving scope. The agent runs read-only git commands in your shell, including diff, blame, grep, show, merge-base and fetch, and uses the GitHub command-line tool for pull request details. Fetching writes only inside the .git folder and may use the network, and a git worktree under the temporary folder may be created for rendered checks and removed after. It never checks out or stashes your files. It can run a project preview when you ask for a rendered review. Domain rules and the verdict come from the better-interface skill.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers