Naming Audit

Naming audit

A skill for auditing naming conventions across a design system's components and documented patterns. Produces a violation report with specific examples, ambiguity flags, and rename suggestions with rationale. Token names are audited by token-audit, prop names by component-api-validator and Figma variable names by figma-variable-audit; this skill cites their findings rather than repeating them.

Before you begin: verify references

Confirm that every path in this skill's frontmatter references: exists relative to this SKILL.md. If any is missing, stop: the install is incomplete, usually because a flattening installer (for example npx skills install) dropped the repo-root knowledge-notes/ directory. Tell the user to reinstall by a method in 1-INSTALL.md and run verify-install.sh from the install root. Proceed without the references only if the user explicitly says to, and then say in the output that it was produced without the pack's reference material.

Context

Naming is the primary interface between a design system and its consumers. A name is the first piece of information a designer or developer gets about what a component does, what a token means, or how a pattern behaves. Good names are predictable: consumers can guess what a name refers to before they look it up. Bad names require lookup, then clarification, then occasionally a conversation to confirm what was meant.

Naming problems accumulate. A single ambiguous component name is an inconvenience. Twenty ambiguous names spread across a library, with some following one convention and others following three others, is a system that new team members cannot navigate and experienced team members cannot trust.

This audit covers naming for components and any documented patterns. It does not mandate a specific naming convention — it assesses whether the naming is consistent, unambiguous, and fit for its purpose.

Step 0: Identify what you're looking at

Before auditing names, determine what kind of shared UI this is. The library type changes how strict the consistency expectations should be and what recommendations are proportionate.

Classify from codebase signals:

• Design system — Full audit applies. Naming conventions should be documented, consistent, and enforced. Recommend creating a decision record for any undocumented conventions.
• Component library — Naming consistency matters, but the recommendation to "create a decision record" should be scaled down to "write down the convention you're already following, even if it's just a comment in the README." A 5-component library does not need a formal governance artefact.
• Pattern library — Naming conventions often follow the documentation tool's conventions (Fractal's folder numbering, Storybook's story hierarchy). Audit against the tool's conventions as well as internal consistency. Pattern names tend to describe what the pattern shows rather than what it does — flag this only if it creates confusion, not as a blanket violation.
• Utility collection — Naming is the primary interface. Every utility name needs to be unambiguous and predictable because there's no documentation site to fall back on. The audit should be strict on clarity and lenient on formal convention — a utility called clamp-width is better than one called u-cw even if the latter follows a prefix convention.

Include the classification in the report header as "Library type: [Design system / Component library / Pattern library / Utility collection]" and calibrate recommendation weight accordingly.

Step 1: Gather the name inventory

Pull names from the source before asking:

• Component names — the public barrel exports (index.ts / index.js at the package root, or each package's entry point in a monorepo). Exports are the names consumers actually type; internal files aren't. Fall back to component directories if there's no barrel. Record the file and line of each export: that is the finding's evidence
• Pattern names — the docs site or Storybook hierarchy, if documented
• Existing naming convention documentation — README, CONTRIBUTING, ADRs
• Neighbouring reports — a token-audit report (token naming findings) or component-api-validator report (prop naming) if either has been run; cite their IDs in the summary rather than re-auditing

Only ask the user for a list if you can't reach the source, and say which names came from where. If no barrel, component directory or docs hierarchy can be found, stop and ask where the components live rather than auditing file names.

If naming convention documentation exists, assess against it. If it does not, derive the implicit conventions from the existing names and note where they are inconsistent with each other.

Small-system note (fewer than 5 components): A naming audit on a system this size is more of a naming workshop than a compliance audit. The consistency check (Step 2) becomes trivial — with 1–4 components, either every name follows the same convention or the inconsistencies are immediately visible. Focus the audit on purpose clarity and ambiguity flags rather than pattern detection. Offer to capture the convention as a decision record (decision-record) now, while the system is small enough to rename without migration cost.

Step 1b: Derive the dominant convention

If no naming convention documentation exists, don't stop to establish one first. Derive the dominant convention from the inventory — the pattern most names already follow — and show it in the report's convention inventory so the team can confirm or correct it:

• Components: casing, specificity direction (ButtonPrimary vs PrimaryButton), abbreviation use, prefix/suffix rules
• Patterns: whether names describe the user's task (Sign in, Filter a list) or the components involved (Form with validation), and whether the docs tool's hierarchy is followed

Audit against that derived convention, and say it's derived. Where no convention dominates, report that as the finding rather than picking one. At the end, offer to capture the convention as a decision record using the decision-record skill.

Step 2: Assess component naming

Consistency check

Are component names following a consistent convention? Identify which conventions are in use:

• Casing: PascalCase, camelCase, kebab-case, or mixed
• Specificity pattern: general-to-specific (ButtonPrimary) or category-first (NavigationPrimary)
• Abbreviation policy: are abbreviations used, and are they consistent (Btn vs Button, Nav vs Navigation)

List the components that break from the pattern the rest of the library follows — worth aligning when they're next touched. Where the difference might be a deliberate naming decision, say so and ask rather than assuming it's an oversight.

Purpose clarity check

A component name should communicate what the component does without requiring context.

Flag names that:

• Are generic to the point of meaninglessness for what they do: Wrapper, Base, Thing, Item on a component with a specific job. Layout primitives are exempt — Box, Stack, Flex, Grid, Container and Layout are well-understood names for general-purpose layout components
• Describe visual treatment rather than function: BlueCard, LargeText, RoundedButton
• Use internal team jargon: names that would not be understood by someone new to the organisation
• Are ambiguous between similar components: Modal and Dialog in the same system, Tooltip and Popover without clear distinction

For each flagged name: describe the ambiguity and suggest a rename. The rename suggestion should follow the system's established convention and improve rather than just change.

Suffix and prefix conventions

If the system uses suffixes or prefixes to indicate category, variant, or role, are they applied consistently?

Common patterns to check:

• Size suffixes: ButtonSm, ButtonMd, ButtonLg — are sizes applied consistently across components?
• State suffixes: -active, -disabled, -loading — are state names consistent across components?
• Category prefixes: Form-, Nav-, Data- — are prefixes applied to all relevant components?

Flag any component that should have a prefix or suffix based on the system's conventions but does not.