Index of every page on this site: llms.txt
All skills
21st-dev avatar

/21st-ui-review

@15c7436
by 21st Labs21st-dev/skill43 stars
11
Run it yourself

Review existing UI against the project's design context, accessibility expectations, responsive behavior, interaction quality, and high-confidence production rules. Use for UI audits, design QA, frontend review, accessibility review, consistency checks, or requests to find and safely fix visual defects. Trigger for requests such as "review this UI", "audit this page", "find design issues", "check responsiveness", or "fix obvious UX problems". Use 21st-ui-build when the request is primarily to create new UI.

Use this Skill: https://skilld.dev/gh/21st-dev/skill/21st-ui-review

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈132 tokens for metadata: the name and description. ≈413 when used: this file. ≈50 more on demand in 1 file.

Description uses 6.6% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈529 characters, or 6.6%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

Review UI with evidence

Separate deterministic defects from subjective design recommendations.

Workflow

  1. Read .21st/design.json, project instructions, and the relevant source.

  2. Run the deterministic review first:

    21st review <path>
    21st review <path> --json
  3. Inspect the actual component composition and runtime states. Validate claims against code rather than inferring behavior from names.

  4. Group findings into:

    • deterministic defects;
    • design-system drift;
    • accessibility and interaction defects;
    • responsive risks;
    • subjective product/design recommendations.
  5. Prioritize by user impact and include exact file and line attribution.

  6. Use 21st review <path> --fix only for deterministic, low-ambiguity fixes. Never silently redesign, change brand identity, or alter product behavior.

  7. Run relevant project tests after fixes and rerun review to prove resolution.

Review priorities

  • semantic controls and accessible names;
  • keyboard access, focus visibility, and dialog/menu behavior;
  • image alternatives and meaningful labels;
  • touch target size and disabled/loading states;
  • responsive overflow and fixed-width assumptions;
  • reduced-motion handling;
  • hardcoded visual values that conflict with project tokens;
  • duplicate primitives instead of existing components;
  • inconsistent typography, radius, spacing, shadows, and icons;
  • fake data, placeholder copy, or missing empty/error states.

Output

Lead with actionable findings ordered by severity. Keep subjective suggestions clearly labeled. If no issue is proven, say so and state what was inspected.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts2mo3 checks · Risk SAFE
  • Gen Agent Trust Hub2mo

    The skill is functional for UI auditing but possesses an indirect prompt injection surface because it processes untrusted local project files and source code to drive its logic and tool execution.

  • Socket2mo

    No alerts

  • Snyk2mo

    Risk: LOW · No issues

Signed by skilld at 15c7436. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last month.

Steadyupdated 3 months ago

README badge

README badge for 21st-dev/skill/21st-ui-review