Content Generation Checklist for Web Development

Decide which assets you will generate, such as API docs, tutorials, landing pages, and release notes, then define concrete KPIs like signups, demo requests, or docs page engagement. Clear outcomes steer the prompt templates and data you need to extract from the codebase.

List exact repositories, paths, and APIs that content must reference, for example OpenAPI specs, GraphQL schemas, Storybook stories, and code examples directories. Maintain a manifest file in the repo so pipelines know where to pull facts from.

Specify audiences like frontend engineers, backend engineers, and full stack teams, plus expected depth and stack examples such as Next.js, Express, or Go. This informs template variants and example code language selection.

Adopt a prose linter such as Vale with rules for tense, glossary, and inclusive language, alongside markdownlint or remark-lint for formatting. Store rules in the repo so CI enforces tone and structure on every generated artifact.

Use Search Console and Ahrefs to identify developer-intent queries like 'Next.js middleware example' or 'OpenAPI pagination pattern'. Tie each topic to a feature or API surface so content aligns with product capabilities.

Define required frontmatter fields such as title, slug, description, audience, stack tags, and owner team. Validate these fields with JSON Schema so generated markdown or MDX is machine-consumable and consistent.

Use CODEOWNERS to specify who reviews tutorials versus API references, and which content types auto-merge if validations pass. This removes bottlenecks while keeping risky pieces gated by the right maintainers.

Export canonical JSON from OpenAPI via tools like Redocly CLI or openapi-cli, or introspect schemas with GraphQL Inspector. Use these artifacts as the factual backbone for endpoint docs and request-response examples.

Use TypeDoc or JSDoc to extract TypeScript types and docstrings, or godoc for Go. Feed extracted signatures and comments into the generation prompts to ensure parameter names and types are exact.

Pull examples from an /examples or /docs/examples folder and unit tests, then run them with ts-node, Node.js, or pytest to verify they execute. Only include passing snippets in the generated content.

Use conventional commits with semantic-release or Changesets to produce machine-readable release data. Map features and fixes to content sections for release notes, upgrade guides, and migration steps.

Parse package.json engines and peerDependencies, or Docker images and runtime versions, to auto-generate support tables. Keep these matrices in sync with the code to prevent outdated claims in docs.

Run Lighthouse CI or WebPageTest on reference apps, then export metrics like LCP and TTI. Use the numbers to ground landing pages and guides in real performance results.

Query Sentry or Datadog for top error signatures and common misconfigurations, then route them into troubleshooting content. Include stack-specific fixes for React, Node.js, or Nginx where relevant.

Set temperature and top_p to 0, and fix random seeds when the API supports it. Keep model and options pinned in code to avoid drift across runs.

Validate model outputs against a strict JSON Schema that includes frontmatter, section ordering, and code block boundaries. Reject and retry on schema violations to prevent malformed MDX or markdown.

Maintain a library of high quality examples per content type, such as an API page and a Next.js tutorial, and store expected outputs as golden files. Unit test prompts to catch regressions when templates change.

Use Handlebars, EJS, or Jinja templates to compose prompts from extracted facts, examples, and style rules. Keep template variables small and well named to limit unexpected generation behaviors.

Run regex or custom checks to block unsafe output such as secrets or internal hostnames, and apply profanity plus PII filters. Fail the job if restricted content is detected.

Commit prompts and schemas to Git with semantic versioning and embed the commit SHA in generated frontmatter. This enables reproducibility and audit trails for every published artifact.

Provide a Node.js or Python CLI that accepts content type, paths, and flags, then call it from GitHub Actions or GitLab CI. Cache extracted data and model responses to reduce run time.

Apply Vale for style and tone, markdownlint or remark-lint for formatting, and alex for inclusive language. Fail builds on violations to maintain consistency across all generated content.

Use lychee or markdown-link-check to verify internal anchors and external URLs. Cache results and retry transient failures to avoid flaky builds.

Run JavaScript and TypeScript snippets with Node.js or ts-node, and Python with pytest, inside Docker where needed. Pin runtime versions so examples behave identically across environments.

Use axe-core or pa11y in a headless browser to scan preview builds for headings, landmarks, contrast, and alt text. Block publishing if high severity issues are found.

Ensure each page has canonical URLs, Open Graph tags, and JSON-LD using the BlogPosting or TechArticle schema. Validate with the Rich Results Test to prevent schema errors.

Measure Flesch Kincaid and word complexity, then enforce range by audience type. Flag vague phrases and require concrete API names and code references.

Scan generated content for secrets using trufflehog or gitleaks and ensure code samples adhere to license constraints. Record compliance results in CI artifacts for auditing.

Integrate content with Docusaurus, Next.js plus Contentlayer, or Astro, then ensure frontmatter fields map to page templates. Keep routes stable to avoid breaking incoming links.

Use Vercel or Netlify to build previews for each PR and require CODEOWNERS approval for sensitive content types. Include a diff of generated frontmatter and structured data in the PR body.

Configure experiments with GrowthBook or Optimizely to test developer focused messaging, for example 'Integrate in 5 minutes' versus 'Type safe SDK in 5 steps'. Guard experiments behind feature flags to limit risk.

Publish to Dev.to, Hashnode, and Medium with canonical links back to your domain. Automate via platform APIs and schedule via CI cron jobs to keep cadence steady.

Add UTM parameters to internal and external links, then monitor conversions in GA4 or Plausible for events like 'Docs copy example' or 'Start trial'. Use a shared naming convention for accurate attribution.

Create dashboards for bounce rate, time on page, and CTA clicks per content type, then alert owners when KPIs drop. Tie alerts to Slack with thresholds so teams respond quickly.

Run openapi-diff or GraphQL Inspector nightly to detect breaking changes that invalidate docs or tutorials, then open issues assigning owners. Include a freshness date in frontmatter and enforce refresh SLAs.