[nehagup]

Fixes the keploy/blog-website items from the Oct 2026 SEO/GEO audit.

Titles and H1s

• <title> clipping (utils/seo.ts): over-long titles now cut at a phrase or word boundary. A clipped title never ends in a connector (&, and, vs, :, ,) or an unclosed (.
  Example: "Integration Testing: Complete Guide (Types, Tools &" becomes "Integration Testing: Complete Guide".
• H1 casing (components/post-title.tsx, new utils/titleCase.ts): the old helper lowercased the whole title and then capitalized after every word boundary. The new helper keeps acronyms (API, QA, AI, CI/CD, SDK, JSON, YAML, UI/UX, HTTP, REST, gRPC, E2E, eBPF and others), keeps mixed-case words as written (JavaScript, GraphQL), does not capitalize after an apostrophe, and keeps "vs" lowercase.
  Examples: /blog/community/what-is-api-testing, /playwright-api-testing, /qa-testing-tools, /what-is-api-mocking ("A Developer'S Guide To Api Mocking" becomes "A Developer's Guide To API Mocking").

Author E-E-A-T

• Empty author box: WordPress (PublishPress) injected an empty author box with the hardcoded box-post-id-16 into every post body. It is now removed by stripPublishPressAuthorBox (lib/author-box.ts). The page's own per-post AuthorCard stays.
• Writer bio and avatar: when the post content has no author-box bio, the writer bio and avatar now come from the author's own WordPress user record (Biographical Info and avatar). The record is fetched by exact display-name match in getStaticProps and memoized. It falls back to the old behaviour on any failure.
• Updated date: the byline now shows "Updated " when WordPress modified is a later day than date. Both dates use <time datetime>.

Schema

• The BlogPosting/TechArticle publisher (@id https://keploy.io/#organization) now carries url and sameAs. sameAs includes https://www.wikidata.org/wiki/Q137395272, which was also added to the global Organization sameAs.
• The mentions SoftwareApplication reference (@id https://keploy.io/#software) now includes applicationCategory: "DeveloperApplication".
• Unchanged, already correct: Person author with @id/url/worksFor, and dateModified, which already comes from the WP modified field.

Pre-content clutter

• The byline, date, tags, "Summarize with AI" and preferred-source row are now wrapped in <aside aria-label="Post details"> instead of a <div>. Same classes, so the visual layout does not change. Extractors should now treat this as page chrome, so the first content passage is the article lede.

Crawl files

• Deleted public/robots.txt (/blog/robots.txt) and its Cache-Control header rule. Crawlers only read /robots.txt at the host root (RFC 9309), so this file was ignored. It also contradicted the root policy on training bots. All rules must live in the root robots.txt (keploy/landing PR). Before merging, make sure the root file has the /blog/api/, /blog/wp/ and /blog/cgi-bin/ disallows (and the /blog/api/proxy-image allow) that this file had.
• Sitemap (scripts/generate-sitemap.mjs): <priority> is no longer emitted (Google ignores it). public/sitemap.xml updates on the next Sync Sitemap run.

Skipped and why

• Blog llms.txt (Allen Digital, "Feneva", stars): public/llms.txt and public/llms-full.txt here contain none of these strings. They are in the root llms.txt served by keploy/landing, so that is where they need fixing. Live GitHub count is 18,689, so it should read "18K+".
• Bulk lastmod: the generator already uses each post's real WP modified date, falling back to date. The shared dates (67 posts on 2026-09-09, 62 on 2026-08-18) come from bulk edits in WordPress itself. Fixing them needs an editorial rule in WordPress, not a code change: avoid bulk re-saves and plugin-wide content rewrites, or use a "substantive update" field.
• wp.keploy.io noindex: this is WordPress server config, so it cannot be set from this repo. See manual steps.

Manual steps (WP admin)

1. On wp.keploy.io, stop the front end from being indexed while keeping /graphql and /wp-content/uploads reachable. Either:

   • send X-Robots-Tag: noindex from the server or a plugin on all non-/graphql front-end responses, or
   • 301 front-end URLs (/?p=* and post permalinks) to https://keploy.io/blog/....

   Do not use Settings > Reading > "Discourage search engines" alone. It only changes robots meta, and some plugins also apply it to feeds.

2. Fill in each author's Biographical Info in Users > Profile. The post author card and the Person schema now read it.

3. Optional: turn off PublishPress's automatic author box ("Show author box after content"). This PR strips it, but it is still dead markup in the WordPress output.

Tests

• npm run test:unit: 135/135 pass. Adds tests/lib/titleCase.test.ts, which covers headline case, phrase-boundary clipping, the author-box strip and the exact-name WP user match.
• npx tsc --noEmit: pass.
• npm run lint: 0 errors (only the existing exhaustive-deps warnings).
• npm run build: not verified locally. It failed inside next/font (Google DM_Sans in pages/_app.tsx, a file this PR doesn't touch) with "Cannot read properties of null", on Node 26.3. That looks like a local Node/Next mismatch, so please rely on the Vercel preview build.

🤖 Generated with Claude Code

[Copilot]

🟡 Changes recommended

Writer-data serialization can expose email addresses, while author-box removal and title processing can corrupt published content.

5 open issues

• Prevent author email exposure in serialized page data · New
• Remove only PublishPress author blocks · New
• Preserve plus signs in programming language names · New
• Track unmatched parentheses when clipping titles · New
• Use a prototype-free acronym dictionary · New

What changed in this PR

Addresses the blog’s SEO/GEO audit with improved title formatting, richer author metadata and structured data, and crawl-file cleanup.

Changes:

• Adds acronym-aware headline casing and phrase-boundary title clipping.
• Adds WordPress author-profile fallbacks, visible updated dates, and author-box cleanup.
• Enriches schema metadata, removes sitemap priorities, and delegates robots rules to the host root.

File	Description
utils/​titleCase.ts	Adds headline casing that preserves acronyms and mixed case.
utils/​seo.ts	Adds boundary-aware title clipping.
tests/​lib/​titleCase.test.ts	Tests casing, clipping, author-box removal, and author matching.
scripts/​generate-sitemap.mjs	Stops emitting sitemap priorities.
README.md	Documents root-level robots policy ownership.
public/​robots.txt	Deletes the blog-local robots file.
pages/​technology/​[slug].tsx	Integrates writer profiles, modified dates, and content cleanup.
pages/​community/​[slug].tsx	Integrates writer profiles, modified dates, and content cleanup.
next.config.js	Removes the blog robots cache-header rule.
lib/​structured-data.ts	Enriches publisher and software entity metadata.
lib/​author-box.ts	Adds author-box stripping and exact-name user matching.
components/​PostHeaderAuthors.tsx	Displays semantic publication and updated dates.
components/​post-title.tsx	Uses the new headline-casing helper.
components/​post-header.tsx	Passes modified dates and groups post details in an aside.

_{Review effort: Balanced}

[Copilot]

🔵 Needs a closer look

Title-formatting issues remain unresolved, and root crawl-rule coordination and a successful preview build still need confirmation.

No open issues 🎉

2 previously missed issues

In code that hasn't changed since last review

Preserve UI/UX acronym casing in headline conversion

utils/​titleCase.ts:12

The canonical forms include CI/CD but omit UI/UX. Because the helper splits tokens only on whitespace, toHeadlineCase("ui/ux testing") returns Ui/ux Testing rather than UI/UX Testing. Add UI/UX to the lookup so this acronym receives the same normalization as CI/CD, and cover the lowercase input in the acronym test.

Handle punctuation when preserving VS Code casing

utils/​titleCase.ts:34

The lookahead compares the entire next whitespace-delimited token, so punctuation defeats the VS Code exception. Debugging in VS Code: A Guide becomes Debugging In vs Code: A Guide, and Debugging in (VS Code) becomes Debugging In (vs Code), lowercasing an already-correct product name in the H1. Strip the next token's surrounding punctuation before comparing it with code, and add regression cases for the colon and closing parenthesis.

5 resolved since last review

• Prevent author email exposure in serialized page data
• Use a prototype-free acronym dictionary
• Track unmatched parentheses when clipping titles
• Preserve plus signs in programming language names
• Remove only PublishPress author blocks

_{Review effort: Balanced}