Cicd Integration

CI/CD Integration

A skill for generating pipeline configurations that automate the quality checks Design System Ops skills perform manually. This bridges the gap between "here are the problems the audit found" and "these problems can never recur because the pipeline catches them."

Output type: File creation. This skill produces pipeline configuration files (YAML), scripts, and documentation. It does not execute pipelines — it generates the configuration that teams add to their repository.

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.

Why this exists

Every audit skill in Design System Ops finds problems. Some of those problems should never have reached a human reviewer because they are mechanically detectable: a hardcoded hex value in a component file, a token alias that references a non-existent token, a component export missing from the barrel file, an accessibility violation that an automated scanner would catch.

CI/CD Integration converts audit findings into automated pipeline checks. The goal is not to replace the audit skills — those handle nuance, context, and cross-skill synthesis that pipelines cannot. The goal is to automate the mechanical subset so that audits focus on the problems only humans can evaluate.

Configuration

If .ds-ops-config.yml exists, follow the configuration-and-recurring knowledge note (../../knowledge-notes/configuration-and-recurring.md). This skill reads:

yamlCopy code
cicd:
  platform: "github-actions"         # github-actions, gitlab-ci, circleci, bitbucket
  package_manager: "npm"              # npm, yarn, pnpm
  node_version: "22"                  # Node.js version (22 or 24; both LTS)
  test_framework: "jest"              # jest, vitest, playwright
  component_library_path: "packages/components"
  token_path: "packages/tokens"
  monorepo: true                      # Whether the project uses a monorepo structure
  triggers:
    - "push to main"
    - "pull request to main"

If no configuration exists, ask for:

1. CI/CD platform (default: GitHub Actions)
2. Package manager (default: npm)
3. Whether the project is a monorepo
4. Paths to token files and component files

Step 0: Assess what to automate

Before generating any pipeline configuration, determine which checks are worth automating. Not everything should be in CI.

Automation decision matrix

Rule of thumb

If the skill's finding includes a specific, unambiguous rule (e.g., "tokens must use kebab-case", "no hex values outside token files"), it can be automated. If the finding requires judgment (e.g., "this token naming could be clearer"), it cannot.

Step 1: Map audit findings to pipeline checks

For each audit finding category, determine the automated check:

Token checks

Component checks

Documentation checks

Step 2: Generate the pipeline configuration

GitHub Actions

Produce a workflow file: .github/workflows/design-system-checks.yml

Structure:

yamlCopy code
name: Design System Quality Checks

# No `paths:` filter here: if these jobs are required checks, a PR that
# doesn't touch the paths would leave them pending forever. The `changes`
# job decides what to run instead; skipped jobs count as passing.
on:
  push:
    branches: [main]
  pull_request:
    branches: [main]

permissions:
  contents: read
  pull-requests: read   # lets the path filter list PR files

concurrency:
  group: ds-checks-${{ github.ref }}
  cancel-in-progress: true

jobs:
  changes:
    runs-on: ubuntu-latest
    outputs:
      tokens: ${{ steps.filter.outputs.tokens }}
      components: ${{ steps.filter.outputs.components }}
    steps:
      - uses: actions/checkout@v7
      - uses: dorny/paths-filter@v4
        id: filter
        with:
          filters: |
            tokens:
              - 'packages/tokens/**'
            components:
              - 'packages/components/**'
              - 'packages/tokens/**'

  token-validation:
    name: Token Validation
    needs: changes
    if: needs.changes.outputs.tokens == 'true'
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v7
      - uses: actions/setup-node@v7
        with:
          node-version: '22'
          cache: 'npm'
      - run: npm ci
      - name: Validate token naming
        run: node scripts/ds-checks/validate-token-names.js
      - name: Validate DTCG format and references
        run: npx terrazzo lint            # or: npx style-dictionary build --config tokens.config.js
      - name: Detect orphaned tokens (warn only)
        run: node scripts/ds-checks/find-orphaned-tokens.js

  component-validation:
    name: Component Validation
    needs: [changes, token-validation]
    if: ${{ !failure() && !cancelled() && needs.changes.outputs.components == 'true' }}
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v7
      - uses: actions/setup-node@v7
        with:
          node-version: '22'
          cache: 'npm'
      - run: npm ci
      - name: TypeScript compilation
        run: npx tsc --noEmit
      - name: Hardcoded values in styles
        run: npx stylelint "**/*.{css,scss}" --ignore-path .gitignore   # declaration-strict-value config from governance-encoder
      - name: Check export completeness
        run: node scripts/ds-checks/verify-exports.js
      - name: Accessibility scan
        run: npm run test:a11y      # prerequisite — see Step 3
      - name: Bundle size check
        run: npx size-limit         # prerequisite — see Step 3

  documentation-validation:
    name: Documentation Validation
    needs: changes
    if: needs.changes.outputs.components == 'true'
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v7
      - uses: actions/setup-node@v7
        with:
          node-version: '22'
          cache: 'npm'
      - run: npm ci
      - name: Check component docs exist
        run: node scripts/ds-checks/check-docs-exist.js
      - name: Verify Storybook stories exist
        run: node scripts/ds-checks/check-stories-exist.js

  visual-regression:
    name: Visual Regression
    needs: changes
    if: github.event_name == 'pull_request' && needs.changes.outputs.components == 'true'
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v7
        with:
          fetch-depth: 0          # Chromatic needs git history for baselines
      - uses: actions/setup-node@v7
        with:
          node-version: '22'
          cache: 'npm'
      - run: npm ci
      - name: Build Storybook
        run: npx storybook build --output-dir storybook-static
      - name: Run visual regression
        # Reuse the build above rather than letting Chromatic build Storybook again
        run: npx chromatic --project-token=${{ secrets.CHROMATIC_PROJECT_TOKEN }} --storybook-build-dir=storybook-static

Pin each action to its current major when you generate the file; the majors above were current when this skill was written. Use the Node version from cicd.node_version (22 or 24). dorny/paths-filter is a third-party action; if the team's policy disallows those, replace the changes job with a git diff --name-only step that sets the same outputs.

Blocking merges needs branch protection. A failing job only blocks a merge if the repository requires it. Tell the user to add the job names as required status checks under branch protection or a ruleset on main; the pipeline can't enforce this itself.

GitLab CI, CircleCI, Bitbucket Pipelines

Generate the same job list as the GitHub workflow (token validation, component validation with the Stylelint step, documentation validation, visual regression on pull requests) in the platform's syntax; don't reduce the check list to fit a shorter template. Platform notes:

• GitLab CI (.gitlab-ci.yml): default.image: node:22, an npm cache keyed on package-lock.json, stages: [validate, visual], and rules with changes: on the token and component paths for merge-request pipelines. Set CHROMATIC_PROJECT_TOKEN as a masked variable. To block merges, enable "Pipelines must succeed" in the merge request settings.
• CircleCI (.circleci/config.yml): a node executor on cimg/node:22.x, an install command with restore_cache/save_cache on package-lock.json, component-validation requiring token-validation. CircleCI has no built-in path filtering; add the path-filtering orb only if run time matters. Block merges with required status checks in the git host.
• Bitbucket Pipelines (bitbucket-pipelines.yml): image: node:22, step definitions reused under pipelines.pull-requests and pipelines.branches.main, condition: changesets: includePaths: to scope by path. Blocking merges needs a merge check requiring passing builds (a Premium feature on Bitbucket Cloud).

Step 3: Generate the helper scripts

Each pipeline check references a script. Generate the scripts in scripts/ds-checks/:

validate-token-names.js

javascriptCopy code
/**
 * Validates token names against the configured naming convention.
 * Exits with code 1 if violations are found.
 *
 * Generated by Design System Ops — cicd-integration
 */

const fs = require('fs');
const path = require('path');

// Configuration — adjust these to match your conventions
// A segment is lowercase letters and digits, optionally hyphenated:
// spacing.4, font-size.base, color.blue.500, color.action.primary-hover,
// button.background.hover all pass. 2–5 segments.
const SEGMENT = '[a-z0-9]+(?:-[a-z0-9]+)*';
const TOKEN_NAME = new RegExp(`^${SEGMENT}(?:\\.${SEGMENT}){1,4}$`);

// Tier comes from where a token is defined (file or set), not from how many
// segments its name has — spacing.4 and button.background.default can't be
// told apart by shape alone.
const TIER_BY_PATH = {
  primitive: 'primitives/',
  semantic: 'semantic/',
  component: 'component/',
};

// Token file path — adjust to your project
const TOKEN_DIR = process.env.TOKEN_DIR || 'packages/tokens/src';

// [Full implementation: recursively read token files, assign each token
//  a tier from TIER_BY_PATH, validate its name against TOKEN_NAME and any
//  tier-specific rules, collect violations, read the threshold for
//  `token-naming` from .ds-ops/quality-gates.yml, output violations, and
//  exit 1 only if the count exceeds the threshold and block_merge is true]

The block above is the header and configuration only; the generated file contains the implementation its comment describes. Provide complete, working implementations for each script. Include:

• Clear comments explaining what the script checks
• Configurable paths and patterns at the top of each file
• Thresholds read from .ds-ops/quality-gates.yml (Step 4) via a shared helper, not hardcoded
• Exit codes: 1 only when a gate with block_merge: true is exceeded; otherwise 0, printing warnings (on GitHub, as ::warning:: lines so they show on the PR)
• Human-readable output: which files, which violations, what to fix
• Provenance comment: "Generated by Design System Ops — cicd-integration"

Scripts to generate

1. validate-token-names.js — Regex-based token name validation (Style Dictionary doesn't validate names; Terrazzo's lint rules can, if the team is on Terrazzo, in which case skip this script)
2. find-orphaned-tokens.js — Cross-reference token definitions with component usage
3. verify-exports.js — Compare directory listing with barrel file exports
4. check-docs-exist.js — Verify documentation files exist for each component
5. check-stories-exist.js — Verify Storybook story files exist for each component
6. read-gates.js — Shared helper that loads .ds-ops/quality-gates.yml and returns the threshold and block_merge flag for a gate

DTCG validation and the hardcoded-value check are not scripts: they run Terrazzo or Style Dictionary, and Stylelint or ESLint, with the config the team has (or governance-encoder writes). Don't reimplement a linter in scripts/ds-checks/.

Prerequisites the pipeline assumes

Two steps call tools the scripts above don't create. Generate them, or list them as prerequisites the team must add before the pipeline goes green:

• Stylelint config with stylelint-declaration-strict-value on the token-backed properties (and eslint-plugin-tailwindcss for Tailwind). Run governance-encoder to write it from the team's rules, or generate a minimal one here and say it's a starting point.
• test:a11y script in package.json — for example "test:a11y": "vitest run --project a11y" with jest-axe/vitest-axe tests, or "test:a11y": "test-storybook" with the Storybook test-runner and axe. Generate one example test per component type found.
• size-limit config — a .size-limit.json listing each entry point with a limit, plus size-limit and its preset (e.g. @size-limit/preset-small-lib) as dev dependencies.

Step 4: Generate a quality gate configuration

Produce a quality gate definition that the pipeline enforces on pull requests:

The scripts/ds-checks/ scripts read this file through read-gates.js. Checks run by other tools (Style Dictionary, axe, size-limit, Chromatic) are gated by their own exit codes and config — set their thresholds there, and say so in PIPELINE.md.

yamlCopy code
# .ds-ops/quality-gates.yml
# Quality gates for design system pull requests, read by scripts/ds-checks/
# Adjust thresholds based on your system's maturity

gates:
  token-naming:
    threshold: 0          # Zero tolerance for naming violations
    block_merge: true
    message: "Token naming violations must be fixed before merge"

  orphaned-tokens:
    threshold: 5          # Allow some orphans during migration periods
    block_merge: false    # Warn only: tokens consumed outside this repo look orphaned
    message: "Possibly orphaned tokens — check external consumers before removing"

  exports:
    threshold: 0
    block_merge: true
    message: "Every component directory must be exported from the package entry point"

  docs-and-stories:
    threshold: 0
    block_merge: false
    message: "Components without docs or stories detected"

Step 5: Generate documentation

Produce a PIPELINE.md file that explains:

1. What each check does — in plain language a product manager would understand
2. How to add a new check — step-by-step for an engineer
3. How to adjust thresholds — where the configuration lives and what each threshold means
4. What to do when a check fails — troubleshooting guide for each check type
5. How this relates to Design System Ops audits — which audit findings each check automates

Adaptation by platform

For teams using Figma MCP

Add a step that auto-pulls Figma variable values and compares them against token file definitions. This catches design-code drift at the CI level.

For teams using Style Dictionary or Terrazzo

Style Dictionary's build catches broken and circular references and, with usesDtcg, bad value shapes; it does not validate names, so validate-token-names.js stays. Terrazzo's lint covers DTCG validation and has configurable lint rules that can replace the naming script. Use the tool the repo has; don't add a second token toolchain for CI.

For teams using Storybook

Integrate the visual regression step with Storybook's built-in visual testing or Chromatic. Generate test-runner configuration for accessibility checks within stories.

For monorepo projects

Scope jobs by changed paths inside the pipeline (the changes job above), not with workflow-level paths: filters, so required checks never sit pending. Token changes run token checks; component changes run component checks. Use job dependencies so that token validation runs before component validation (since components depend on tokens).

Quality checks

• Pipeline configuration is valid YAML that passes the platform's schema validation
• All referenced scripts exist and are executable, and every tool a step calls (test:a11y, size-limit) is either generated or listed as a prerequisite
• Actions are pinned to current majors and Node is 22 or 24
• Gate thresholds come from .ds-ops/quality-gates.yml, and the user is told to enable branch protection for blocking checks
• Scripts have clear error messages that explain what went wrong and how to fix it
• Quality gate thresholds are reasonable defaults (not so strict they block everything, not so loose they catch nothing)
• Documentation explains every check in language a non-engineer can understand
• Provenance marker present in all generated files
• Pipeline configuration respects monorepo structure if applicable
• No secrets or credentials are hardcoded — all sensitive values use environment variables or secrets management