What is the impeccable audit skill?
Impeccable audit is a command of the impeccable skill by Paul Bakaus. It runs a technical audit across five dimensions, scores each from 0 to 4 for a total out of 20, tags issues P0 to P3, and writes a report. It documents problems and does not fix them.
Impeccable audit is one command inside the impeccable skill, which the project describes as design guidance for AI coding agents. You run it as /impeccable audit plus a target such as a page or component.
It is a technical audit, not a design review. The agent checks what is measurable in the code and the rendered page, then writes a scored report. The project's separate critique command covers taste, hierarchy and design quality.
What changes in your agent's output
- The agent scores five dimensions from 0 to 4: accessibility, performance, theming, responsive design and implementation integrity.
- The total is out of 20 and maps to a band: 18 to 20 excellent, 14 to 17 good, 10 to 13 acceptable, 6 to 9 poor, 0 to 5 critical.
- Each issue is tagged P0 to P3, so you fix blockers first.
- The report ends with recommended actions that route to other impeccable commands, finishing with polish.
Who the impeccable audit skill is for
Best for
- Pages and components you want scored before a release.
- Teams that need a ranked list of accessibility and performance issues in one report.
- Reviews of AI-generated UI, where implementation integrity is the open question.
Not for
- Fixing problems. The command documents issues and leaves the changes to other commands.
- Judging visual taste. Use impeccable critique for hierarchy and design quality.
- Native apps. The web audit routes native targets to a separate file.
Use it with UiChemy
UiChemy builds WordPress pages from design or from AI output. Run the audit on the built page to see where accessibility and responsiveness fall short, then fix them in Composer or in your builder. Click Get Started on uichemy.com.
How to install Impeccable Audit Skill: Technical UI Audit for AI Agents
npx impeccable installRun the installer
From your project root, run
npx impeccable install. It sets up the impeccable skill for the AI coding agents you pick. Flags such as--providersand--scope=projector--scope=globalnarrow the targets.Create project context
Run
/impeccable initonce. It writes a PRODUCT.md that the commands read, so the output matches your audience and brand.Run the command
Open your agent in the project and run
/impeccable auditfollowed by a target, for example/impeccable audit blog. Run/impeccable pin auditif you want a standalone/auditshortcut.
Check that it loaded
Run /impeccable audit on one page. You should get an Audit Health Score out of 20, an issue list tagged P0 to P3, and a closing list of recommended actions. No files in your project should change.
Key features
- Five scored dimensionsAccessibility, performance, theming, responsive design and implementation integrity each get a 0 to 4 score, so weak areas stand out.
- Rating bandsThe /20 total maps to excellent, good, acceptable, poor or critical, so a number turns into a decision.
- P0 to P3 severityEvery issue carries a priority tag. P0 items block release and P3 items are polish.
- Implementation integrity verdictA dedicated section judges whether the code is a coherent implementation or generic AI output.
- Structured reportSections run from health score and executive summary to detailed findings, systemic patterns, positive findings and recommended actions.
- Positive findingsThe report records what already works, so a later fix pass does not undo it.
- Read-only by designThe command documents and never edits. Fixes happen in follow-up commands.
- Bundled detectorThe audit runs the project's deterministic detector on the target as part of the pass.
- Hand-off to other commandsRecommended actions name the impeccable commands to run next and end with polish.
Frequently asked questions
What is impeccable audit?
It is the audit command of the impeccable skill by Paul Bakaus. It makes an AI coding agent run a technical review of a UI across five dimensions, score it out of 20, and report prioritized issues. It is the measurable counterpart to the critique command.
How do I install impeccable audit?
Run npx impeccable install from your project root, then /impeccable init to create PRODUCT.md. Audit is a command inside the impeccable skill, so there is no separate package. Call it with /impeccable audit and a target, or pin it as /audit.
What does the audit score?
Five dimensions, each 0 to 4: accessibility, performance, theming, responsive design and implementation integrity. The total out of 20 falls into a band from critical at 0 to 5 up to excellent at 18 to 20.
Does impeccable audit fix issues?
No. It documents findings and recommends which commands to run next. Fixing is left to other commands, and the report closes by pointing to impeccable polish for the final pass.
What is the difference between audit and critique?
Audit is technical and scored: accessibility, performance, theming, responsiveness and code integrity. Impeccable critique is a design review of hierarchy, clarity and taste. Use audit for what can be measured and critique for how it feels.
Which AI agents can use impeccable audit?
The impeccable project lists support for Claude Code, Cursor, Gemini CLI, Codex CLI, Grok Build, Hermes and Veto among others. The installer takes a providers flag so you choose which agents receive the skill.
Can it audit native apps?
The web audit is for web targets. The command file routes native app targets to a separate audit.native file, so mobile and native projects should be audited through that route instead of the web one.
What should I run after the audit?
Work through P0 and P1 items first, since they block release. For accessibility depth, read fixing-accessibility. For spacing problems, impeccable layout helps. Then run polish for the final pass on the page.