test(dialtone-docs): DLT-3268 refactor test suite — consolidate, remove hardcoded values, add CI workflow

[belumontoya]

test(dialtone-docs): DLT-3268 refactor test suite

Obligatory GIF (super important!)

🛠️ Type Of Change

• Test

📖 Jira Ticket

https://dialpad.atlassian.net/browse/DLT-3268

📖 Description

Consolidates the dialtone-docs test suite from 6 files / 45 tests down to 3 files / 29 tests, and adds a CI workflow.

Files created

• tests/tests/schema.test.js — Merges frontmatter.test.js + build-output.test.js into a single file with shared ALLOWED_TYPES constant. Validates source markdown frontmatter schema (6 tests) and compiled ai-docs.json output schema (11 tests).
• tests/tests/content-quality.test.js — Replaces hardcoded file lists with dynamic discovery. Uses readdirSync to discover category directories at runtime and verify each has an INDEX.md. Also validates no placeholder text (TODO/FIXME/TBD) in prose and that Handlebars templates exist in src/templates/.
• .github/workflows/docs_tests.yml — New CI workflow that runs pnpm nx run dialtone-docs:test on PRs and pushes to staging touching packages/dialtone-docs/**.

Files deleted

• tests/tests/completeness.test.js — Hardcoded file lists per category (broke every time a doc was added/removed)
• tests/tests/frontmatter.test.js — Merged into schema.test.js
• tests/tests/searchability.test.js — Hardcoded content term assertions (not structural)
• tests/tests/structure.test.js — Config file existence checks (near-zero value)
• tests/tests/build-output.test.js — Merged into schema.test.js
• tests/helpers/markdownParser.js — 333 lines of unused helper code (never imported by any test)

Unchanged

• tests/tests/strip-markdown.test.js — 8 unit tests, untouched

💡 Context

The dialtone-docs test suite had scaling problems:

• Hardcoded file lists — tests listed every doc filename per category, requiring manual updates whenever a doc was added or removed
• ALLOWED_TYPES duplicated — defined identically in frontmatter.test.js and build-output.test.js
• Content assertions — tests checked that specific package names and technical terms appeared somewhere in the docs (not structural)
• Dead code — markdownParser.js (333 lines) was never imported by any test
• No CI — tests only ran on command, not on PRs

All remaining tests now validate structure and schema only — they never break when docs content changes.

📝 Checklist

• I have ensured no private Dialpad links or info are in the code or pull request description (Dialtone is a public repo!).
• I have reviewed my changes.
• I have added all relevant documentation.
• I have considered the performance impact of my change.
• I have added / updated unit tests.

🔮 Next Steps

• Wire .claude/skills/, .claude/agents/, .claude/commands/, .claude/hooks/ into the doc-sync-enforcer pipeline so skill/agent changes trigger doc updates

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 27920e2bcb

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "Codex (@codex) address that feedback".

[chatgpt-codex-connector]

Preserve YYYY-MM-DD when serializing lastUpdated

gray-matter (via YAML parsing) materializes unquoted dates like 2026-03-04 as Date objects, so String(frontmatter.last_updated) emits a locale/timezone-dependent value (e.g., Wed Mar 04 2026 ...) instead of the documented YYYY-MM-DD schema. This makes ai-docs.json nondeterministic across environments and can break downstream consumers that expect normalized date strings; serialize to an explicit date format (or preserve the original scalar text) before writing lastUpdated.

Useful? React with 👍 / 👎.