CI for Docusaurus Documentation

Docusaurus is great. Your docs CI isn't.

Docusaurus makes it easy to write beautiful documentation. But it doesn't verify that the code inside actually works — and neither does your default CI setup.

• ✗Code snippets in MDX files become stale as your API evolves — no automated check catches it
• ✗Accessibility issues in custom React components go unnoticed until users report them
• ✗When your OpenAPI spec changes, documented request/response examples silently diverge
• ✗Broken examples in docs cause support tickets and damage developer trust
• ✗Writing a custom test script for Docusaurus MDX fenced code blocks takes days and breaks often

DocsCI: the missing CI layer for Docusaurus

DocsCI reads your Docusaurus repository, extracts every fenced code block from MDX and Markdown files, and executes them in sandboxed runtimes. It detects API drift between your OpenAPI spec and documented examples, checks accessibility in rendered components, and posts precise PR comments with AI-generated fixes. Add one GitHub Action — no changes to your Docusaurus config.

Everything you need

How it works

# .github/workflows/docsci.yml
- name: Run DocsCI
  run: |
    curl -sf -X POST https://snippetci.com/api/runs/queue \
      -H "Authorization: Bearer ${{ secrets.DOCSCI_TOKEN }}" \
      -F "docs_archive=@docs.tar.gz"

# docsci.yml (optional)
docs:
  path: ./docs
  include: ["**/*.md", "**/*.mdx"]
openapi:
  url: https://api.example.com/openapi.json
checks:
  snippets: true
  accessibility: true
  drift: true

Frequently asked questions