[HardeepAsrani]

Summary

When the page title is hidden, views/page-header.php printed a screen-reader H1 unless it found an <h1 in the post's raw post_content. Anything else that renders an H1 was invisible to that check, so the page ended up with two: Neve Pro custom layouts on hooks or inside the content, synced patterns, page builders that store markup in meta, and every non-singular page (the blog page hides its title by default for sites installed after 4.0).

The decision is now deferred. Page_Header opens an output buffer on the theme's first body hook, neve_body_start_after (with wp_body_open as a fallback for other templates), when the title is hidden. The view prints a placeholder comment where the title would be, and the buffer callback swaps it for the screen-reader H1 only if the buffered body contains no H1. The scan is a single linear pass, so a large inline script cannot hit the PCRE backtrack limit; it ignores HTML comments, the raw-text elements (script, style, noscript, textarea, title, xmp, iframe, noembed, noframes), tag text inside quoted attribute values, and anything inside a <template>, whose contents nest like regular markup. Inside <svg> and <math> it follows the parser's foreign-content rules: no raw text, self-closing tags, and an <h1> that breaks out as HTML, with the HTML rules restored inside integration points such as <foreignObject>, SVG <title> and <desc> and the MathML text elements. Script and style end tags that carry attributes still close the element, as in browsers. The buffer closes on the last body hook, neve_body_end_before, so every custom layout hook from body start to body end counts; if that hook never fires, PHP runs the callback at shutdown. When no buffer is active (a template without the body hooks), the previous post_content scan still applies.

Reproduced on a local site with Neve Pro: hook layout before the loop, hook layout before content, inside-content layout, synced pattern, and the blog page. All produced an extra H1 before and none after. Beaver Builder and Brizy were not available locally; the hooked-H1 unit tests cover the same mechanism.

Will affect visual aspect of the product

NO. The fallback H1 is visually hidden; only its presence changes.

Test instructions

• Hide a page's title (page settings or Customizer) and give it no H1: exactly one h1.screen-reader-text renders.
• Same page with an H1 block in the content: no screen-reader H1.
• Put the H1 in a synced pattern instead: no screen-reader H1.
• With Neve Pro, add a Custom Layout containing an H1 on neve_before_content, neve_after_header_hook or neve_before_footer_hook, or an inside-content layout: no screen-reader H1.
• Set a static front page and a posts page; visit the posts page with a Custom Layout H1 on neve_before_loop: one H1.
• Leave the title visible: the normal <h1> renders and no placeholder comment appears in the source.

Check before Pull Request is ready:

• I have written a test and included it in this PR — tests/test-neve-hidden-title-h1.php (62 cases: every body hook, the stray-buffer guard, a 1.2 MB inline script, and a provider of comment, raw-text, nested-template and attribute edge cases; the behavior cases fail on the base tree) and a synced-pattern fixture plus assertion in e2e-tests/specs/a11y-ready/structure.spec.ts.
• I have run all tests and they pass — PHPUnit 154 tests; the three hidden-title e2e tests pass against a local site.
• The code passes when running the PHP CodeSniffer
• Code meets WordPress Coding Standards for PHP, HTML, CSS and JS
• Security and Sanitization requirements have been followed — title still passes through wp_kses_post(); buffered template output is passed through unchanged.
• I have assigned a reviewer or two to review this PR

Review follow-up

The automated review asked whether the test-name conventions in CONTRIBUTING.md (lines 246 and 260-263) should be updated, since no existing test follows them. That predates this change and applies to the whole suite, so it is left for a separate docs PR rather than mixed into this fix.

Closes #4652.

🤖 Generated with Claude Code

[pirate-bot]

Plugin build for 0fe4373 is ready 🛎️!

• Download build

[pirate-bot]

🟡 Changes requested

The fix scans only the main area, so a custom layout H1 in the header or footer still gives two H1s. The H1 check, three tests and one type also need fixes.

Validation details

• Files reviewed: 5/5 changed files. Unchecked history rows: Footer Menu duplicates footer-menu ID across responsive footer layouts, Page title disappears on post category pages, Accessibility changes to match EU accessibility requirements (EN 301 549) that become mandatory in June 2025, Submenu toggle does not open through screen-reader activation, Page Title visible, even when it's disabled from Neve options, Issue with archive title, Featured post section enabled makes page titles disappear, Pass TA guidelines.
• The PR's PHPUnit class passes on HEAD with PHP 8.3 and WordPress 7.1.3: mkdir -p /workspace/checks/R/head && cd /workspace/checks/R/head && rm -rf neve && cp -a /repo neve && cd neve && set -o… printed OK (8 tests, 16 assertions) (run R2).
• The same class on the base tree: cd /workspace/checks/R && rm -rf basetree && mkdir basetree && cp -a /repo basetree/neve && cd basetree/neve && for f in… printed Tests: 8, Assertions: 11, Errors: 2, Failures: 2. (run R4) Only the two hooked-H1 tests fail there.
• An H1 on a header or footer hook still duplicates: `mkdir -p /workspace/checks/K; cat > /workspace/checks/K/test-zz-review.php <<'EOF'

Untested areas

- The e2e spec did not run. The sandbox has no built assets or served site. - Neve Pro was not available. Its custom layout rendering was not exercised. - These history rows were not re-verified by a dedicated check: Footer Menu duplicates `footer-menu` ID across responsive footer layouts, Page title disappears on post category pages, Accessibility changes to match EU accessibility requirements (EN 301 549) that become mandatory in June 2025, Submenu toggle does not open through screen-reader activation. These rows were checked only with ad-hoc unit runs: Page Title visible, even when it's disabled from Neve options, Issue with archive title, Featured post section enabled makes page titles disappear, Pass TA guidelines.

🤖 Automated review · run code-review-agent_6ad063fb8f1ba9.24760594. --- **🤖 Review agent** — review posted ✅ on `1f278ccd` · [changes requested · 5 findings](https://github.com//pull/4658#pullrequestreview-5528869504) · 39 min Run `code-review-agent_6ad063fb8f1ba9.24760594` · [trail](https://agents.vertistudio.com/admin/trails/github-pr/Codeinwp-neve-4658)

[pirate-bot]

🟡 Changes requested

The earlier findings are fixed, but an H1 on the body start or body end hook still gets a second H1. Fix that, one stale comment and two test issues.

Validation details

• Files reviewed: 5/5 changed files. Unchecked rows: Optional setting off: Producer, Setting unset, Override, Stale data.
• The PR's PHPUnit class passes on HEAD with PHP 8.3 and WordPress 7.1.3: grep -E "^[0-9]+\) |Error: |Tests:" /tmp/sandbox-phpunit-ab.zVTgcX/base.log; mkdir -p /workspace/checks/R/head && cd /wo… printed OK (10 tests, 26 assertions) (run R1).
• On the base tree, eight tests fail on their assertions: cd /workspace/checks/R && rm -rf basetree && mkdir basetree && cp -a /repo basetree/neve && cd basetree/neve && for f in… printed Tests: 10, Assertions: 9, Errors: 2, Failures: 8. (run R2)
• Real template renders give one H1 for header and footer hooks: cd /workspace/checks/R/head/neve && phpunit --do-not-cache-result --filter 'TestZzReview::test_real_page with data set "… printed [after_header_hook] h1=1 sr=0 (run R20), cd /workspace/checks/R/head/neve && phpunit --do-not-cache-result --filter 'TestZzReview::test_real_page with data set "… printed [before_footer_hook] h1=1 sr=0 (run R21).
• The body start hook still duplicates: cd /workspace/checks/R/head/neve && phpunit --do-not-cache-result --filter 'TestZzReview::test_real_page with data set "… printed [body_start_after] h1=2 sr=1 (run R15).

Untested areas

• The e2e spec did not run. The theme needs built assets before a served site loads it.
• Neve Pro was not available. Its custom layout hook list was not read.
• test_unbuffered_path_keeps_content_scan errors on the base tree on a missing method. Its base result is inconclusive.
• These setting rows were not settled: Optional setting off: Producer, Setting unset, Override, Stale data.

🤖 Automated review · run code-review-agent_6ad1ce22a74bb3.68194372.

---

🤖 Review agent — review posted ✅ on 937dfe56 · changes requested · 4 findings · 46 min

Run code-review-agent_6ad1ce22a74bb3.68194372 · trail

[pirate-bot]

🟡 Changes requested

The earlier findings are fixed. Three test issues remain: one order-dependent test, one weak assertion and one duplicated helper body.

Validation details

• Files reviewed: 5/5 changed files. Unchecked rows: Optional setting off: Producer, Setting unset, Override, Empty data.
• The PHPUnit class ran on HEAD with PHP 8.3 and WordPress test library mkdir -p /workspace/checks/R/head /workspace/checks/R/html /workspace/checks/T && cd /workspace/checks/R/head && rm -rf … printed Tests: 14, Assertions: 36, Failures: 1. (run R2)
• With kses_remove_filters() added, it passes cd /workspace/checks/R/head/neve && cp /workspace/checks/R/orig-test.php tests/test-neve-hidden-title-h1.php && sed -i '… printed OK (14 tests, 38 assertions) (run R6)
• On the base tree, twelve tests fail on assertions cd /workspace/checks/R && rm -rf base && mkdir base && cp -a head/neve base/neve && cp orig-test.php base/neve/tests/tes… printed Tests: 14, Assertions: 14, Errors: 2, Failures: 12. (run R14)
• Real template renders give one H1 for the body start hook cd /workspace/checks/R/head/neve && for ds in body_start_after body_end_before after_header_hook before_footer_hook nest… printed [body_start_after] h1=1 sr=0 (run R11)

Untested areas

• The e2e spec did not run. The theme needs built assets.
• Neve Pro was not available. Inside-content layouts were not checked.
• Two tests error on the base tree on missing symbols. Their base result is inconclusive.
• These setting rows were not settled: Optional setting off: Producer, Setting unset, Override, Empty data.

🤖 Automated review · run code-review-agent_6ad1db014396c8.21985341.

---

🤖 Review agent — review posted ✅ on 121d8f84 · changes requested · 3 findings · 27 min

Run code-review-agent_6ad1db014396c8.21985341 · trail

[pirate-bot]

🟢 Approved

The PR passes. The three earlier test findings are fixed, and the new commit adds no problem.

Validation details

• Files reviewed: 5/5 changed files. Unchecked rows: Optional setting off: Producer, Optional setting off: Setting unset, Optional setting off: Override.
• The hidden-title test class passes alone on HEAD with PHP 8.3: mkdir -p /workspace/checks/R1/head; cp -a /repo /workspace/checks/R1/head/neve; python3 /workspace/checks/R1/record.py R… printed OK (14 tests, 38 assertions) (run R1)
• The stray-buffer test fails when the buffer-level guard is deleted: set -o pipefail; phpunit --do-not-cache-result --filter 'TestNeveHiddenTitleH1::test_stray_open_buffer_is_left_alone' 2>… printed Failed asserting that 3 is identical to 4. (run R2)
• Both modified tests fail on the base tree: python3 /workspace/checks/R1/record.py R103 /workspace/checks/R1/head/neve "sandbox-phpunit-ab.sh /workspace/checks/R1/h… printed Failed asserting that 2 is identical to 1. (run R3)
• Product-file checks are reused from run code-review-agent_6ad1db014396c8.21985341, not rerun. Those files did not change since then.

Untested areas

• These setting rows were not settled: Optional setting off: Producer, Optional setting off: Setting unset, Optional setting off: Override.
• The e2e spec did not run. The theme needs built assets.
• Neve Pro was not available. Inside-content layouts were not checked.

🤖 Automated review · run code-review-agent_6ad2132037a466.47610119.

---

🤖 Review agent — review posted ✅ on 0830d445 · approved · 0 findings · 25 min

Run code-review-agent_6ad2132037a466.47610119 · trail

[pirate-bot]

🟡 Changes requested

The new scanner calls ctype_space(), which can be missing on some PHP builds. Replace it with strspn().

Validation details

• Files reviewed: 5/5 changed files. Unchecked rows: Optional setting off: Producer, Optional setting off: Setting unset, Optional setting off: Override.
• The hidden-title tests pass on HEAD with PHP 8.3: command -v phpunit; phpunit --version 2>&1 | head -2; cd /workspace/checks/R-T1/head/neve && set -o pipefail && timeout … printed OK (28 tests, 54 assertions) (run R3)
• The new tests fail with the previous regex scanner: cd /workspace/checks/R-T1 && rm -rf variant-b && mkdir variant-b && cp -a head/neve variant-b/neve && cd variant-b/neve … printed Failures: 3 (run R4)
• The new and old scanners differ on two inputs only. Both now match browser parsing [R5].
• Hook and render checks are reused from run code-review-agent_6ad2132037a466.47610119, not rerun.

Untested areas

• Optional setting off: Producer, Setting unset, Override. Open inc/views/page_header.php:97-113. Confirm a filtered page_for_posts or theme mod gives the gate and the view the same result.
• Forms like <!--> and </script/> hide a real H1 from the scanner. No producer was found.
• The e2e spec did not run. Neve Pro was not available.

🤖 Automated review · run code-review-agent_6ad5bb7fbe33a0.27973925.

---

🤖 Review agent — review posted ✅ on 6e633c29 · changes requested · 1 finding · 30 min

Run code-review-agent_6ad5bb7fbe33a0.27973925 · trail

[pirate-bot]

🟢 Approved

The PR passes. The ctype_space() finding is fixed with strspn(), and the new commit adds no problem.

Validation details

• Files reviewed: 5/5 changed files. Commit 6e22b17 changes only inc/views/page_header.php.
• An isolated PHP 8.3 run compared the old and new closing-tag scans. Only a vertical tab differs: python3 /workspace/checks/L1/build.py && { printf 'Run recorded: R1 (L1)\nCommand: timeout 30 php /workspace/checks/L1/c… printed warnings/exceptions=0 (run R3)
• The PR's PHPUnit class passes at HEAD: set -o pipefail; rm -rf /workspace/checks/R1/head && mkdir -p /workspace/checks/R1/head/neve && cd /repo && git archive … printed OK (28 tests, 54 assertions) (run R6)
• The earlier checks on unchanged code come from run code-review-agent_6ad5bb7fbe33a0.27973925 and were not rerun.

Untested areas

• The requirements, security and standards passes were not rerun. The new commit changes two scanner lines and no trust boundary.

🤖 Automated review · run code-review-agent_6ad707cff2da17.80540549.

---

🤖 Review agent — review posted ✅ on 6e22b17a · approved · 0 findings · 27 min

Run code-review-agent_6ad707cff2da17.80540549 · trail

[pirate-bot]

🟡 Changes requested

The new template depth counter works, but two new test rows pass without the change. Delete them.

Validation details

• Files reviewed: 2/5 changed files in commit b01881a. The other three are unchanged since the approved review. Unchecked rows: Optional setting off: Producer, Optional setting off: Setting unset, Optional setting off: Override.
• The PR's PHPUnit class passes at HEAD: OK (35 tests, 61 assertions) (R3).
• With the 6e22b17 scanner, four new rows fail: Failures: 4 (R4).
• An isolated PHP 8.3 run checked nested, unterminated and stray template tags: php /workspace/checks/C1a/driver.php (R5).
• Earlier checks on unchanged code come from run code-review-agent_6ad707cff2da17.80540549 and were not rerun.

Untested areas

• Optional setting off: Producer, Optional setting off: Setting unset and Optional setting off: Override were not rechecked. The commit does not change is_title_hidden() in inc/views/page_header.php.
• The requirements and security passes were not rerun. The commit changes scanner logic and test rows only.

🤖 Automated review · run code-review-agent_6ad72d158c78d6.04717316.

---

🤖 Review agent — review posted ✅ on b01881a8 · changes requested · 1 finding · 16 min

Run code-review-agent_6ad72d158c78d6.04717316 · trail

[pirate-bot]

🟢 Approved

The PR passes. Commit 469eeae deletes the two test rows that passed without the change, and it adds no new problem.

Validation details

• Files reviewed: 1/5 changed files in commit 469eeae. The other four are unchanged since the last review.
• The commit only deletes the template name prefix row and the h1 after template holding a commented closer row in tests/test-neve-hidden-title-h1.php.
• The PHPUnit class passed on b01881a with OK (35 tests, 61 assertions). Source: run code-review-agent_6ad72d158c78d6.04717316, not rerun.
• A source check at HEAD rechecked the producers, defaults, overrides and empty values of the hidden-title settings.

Untested areas

• The PHPUnit class was not rerun at 469eeae. The commit only removes two data-provider rows.

🤖 Automated review · run code-review-agent_6ad74175b801c2.96762909.

---

🤖 Review agent — review posted ✅ on 469eeaea · approved · 0 findings · 18 min

Run code-review-agent_6ad74175b801c2.96762909 · trail

[pirate-bot]

🔵 Needs a closer look

Dev review: The new H1 scan misreads custom-element tags with tag text in quoted attributes. Confirm that no body output carries such markup.

Validation details

• Files reviewed: 2/5 changed files in commit ff60927. The other three are unchanged since the approved review.
• An isolated PHP 8.3 run passed every scanner row at HEAD: timeout 30 php /workspace/checks/L1/check.php printed a: 24/24 pass (run R1)
• The six new rows fail on the previous scanner: timeout 30 php /workspace/checks/L1/check.php printed b: 18/24 pass (run R1)
• WordPress 7.1.3 WP_HTML_Tag_Processor sees an H1 that the scan misses: mkdir -p /workspace/checks/L2a/core cp /var/www/html/wp-includes/html-api/class-wp-html-{tag-processor,attribute-token,d… printed 2 HEAD=false tags=MY-EL,H1 (run R3)
• On a rendered page, wptexturize changed the attribute quote in saved content: cd /var/www/html && wp eval 'echo wptexturize("<div data-tpl=\"<script>\"></div>"), "\n"; echo wptexturize("<my-el data-… printed <my-el data-tpl="<script>&#8220;></my-el> (run R29)
• The settings rows were rechecked from source at HEAD.

Untested areas

• Open inc/views/page_header.php:205. Check whether any plugin prints a hyphenated custom element, such as <my-el data-tpl="<script>">, unescaped into the page body. If none does, the scan is safe. If one does, the page gets a second H1, or none. Then widen the tag name to [a-z][^\s/>]*.
• The block editor UI and REST saves were not exercised.

🤖 Automated review · run code-review-agent_6ad9a0b23053c1.68865080.

---

🤖 Review agent — review posted ✅ on ff60927a · commented · 0 findings

Run code-review-agent_6ad9a0b23053c1.68865080 · trail

[pirate-bot]

🟡 Changes requested

The custom-element fix works, but the widened tag pattern now counts a trailing <h1 as an H1. Restore the tag-name boundary.

Validation details

• Files reviewed: 2/5 changed files in commit 28d2125. The other three are unchanged since the last review. Unchecked rows: Optional setting off: Producer, Optional setting off: Setting unset, Optional setting off: Override.
• An isolated PHP 8.3 scanner run passed every provider row at HEAD: timeout 30 php /workspace/checks/L3a/check.php; printf '\n--- SOURCE LINES ---\n'; nl -ba /workspace/checks/L3a/page_hea… printed HEAD: 26/26 pass; failing rows: none (run R1)
• The two new rows fail on the previous scanner: timeout 30 php /workspace/checks/L3a/check.php; printf '\n--- SOURCE LINES ---\n'; nl -ba /workspace/checks/L3a/page_hea… printed ff60927: 24/26 pass; failing rows: h1 text in custom element attribute; template opener in custom element attribute (run R1)
• Question for the author: the PR body says 39 test cases. HEAD has 41.

Untested areas

• Optional setting off: Producer, Setting unset and Override were not rechecked in this run. Their earlier answers rest on unchanged code in inc/views/page_header.php:98-112.
• The block editor and full page renders were not exercised.

🤖 Automated review · run code-review-agent_6ad9e7a97060a0.96894047.

---

🤖 Review agent — review posted ✅ on 28d21255 · changes requested · 1 finding · 12 min

Run code-review-agent_6ad9e7a97060a0.96894047 · trail

[pirate-bot]

🟢 Approved

The PR passes. Commit 9ee144f fixes the trailing <h1 finding, and it adds no new problem.

Validation details

• Files reviewed: 2/5 changed files in commit 9ee144f. The other three are unchanged since the last review. Unchecked rows: Optional setting off: Producer, Optional setting off: Setting unset, Optional setting off: Override.
• An isolated PHP 8.3 scanner run passed every provider row at HEAD: set -o pipefail; timeout 30 php /workspace/checks/L1a/check.php 2>&1 | tee /workspace/checks/L1a/R1.log; result=${PIPEST… printed SUMMARY 9ee144f: 28/28 pass; failing rows: [] (run R1)
• The two new rows fail on the previous scanner: set -o pipefail; timeout 30 php /workspace/checks/L1a/check.php 2>&1 | tee /workspace/checks/L1a/R1.log; result=${PIPEST… printed 26/28 pass; failing rows: ["h1 opener at end of input","h1 with unterminated attribute"] (run R1)
• A source read of the new comment and test rows found no style or test problem.

Untested areas

• Optional setting off: Producer, Setting unset and Override were not rechecked. Their earlier answers rest on unchanged code in inc/views/page_header.php:98-112.
• The PHPUnit suite, e2e tests and full page renders were not run in this review.

🤖 Automated review · run code-review-agent_6ad9eb5f82f5f3.07096078.

---

🤖 Review agent — review posted ✅ on 9ee144f1 · approved · 0 findings · 6 min

Run code-review-agent_6ad9eb5f82f5f3.07096078 · trail

[pirate-bot]

🟡 Changes requested

The raw-text tag list now misreads a self-closing SVG <title/>, so a later real H1 still gets the fallback H1. Fix the scan.

Validation details

• Files reviewed: 2/5 changed files in commit 55d313c. The other three are unchanged since the last review. Unchecked row: Optional setting off: Producer.
• An isolated PHP 8.3 run of the HEAD and 9ee144f scanners shows each new test row fails on the old scanner [R1].
• The same run shows <svg><title/></svg><h1>Real</h1> returns false at HEAD and true at 9ee144f [R1].
• Source checks found no new security boundary and no stale comment.

Untested areas

• Optional setting off: Producer. The writers of the hide-title settings were not rechecked.
• The YITH gallery page was not rendered in a browser.
• Gutenberg's opt-in full-page navigation closes the top output buffer on wp_footer. Its effect on a later H1 was not run.
• The stripos() cost per raw-text element on PHP below 8.2 was not measured.

🤖 Automated review · run code-review-agent_6ada12780d4c95.45340871.

---

🤖 Review agent — review posted ✅ on 55d313c4 · changes requested · 1 finding · 45 min

Run code-review-agent_6ada12780d4c95.45340871 · trail

[pirate-bot]

🟡 Changes requested

The self-closing SVG <title/> finding is fixed. The new SVG and MathML scan state has no test rows, so add the four rows named inline.

Validation details

• Files reviewed: 2/5 changed files in commit 26fded7. The other three are unchanged since the last review. Unchecked rows: Optional setting off: Producer, Optional setting off: Setting unset, Optional setting off: Override.
• An isolated PHP 8.3 run shows both new rows fail on the 55d313c scanner and pass at HEAD [R2].
• The same run shows four scanner mutants still pass all 33 provider rows python3 /workspace/checks/L1a/build.py && timeout 30 php /workspace/checks/L1a/run.php | tee /workspace/checks/L1a/provi… printed B M1 ALL_PASS=true pass=33/33 failed=[] (run R2).
• A parse5 comparison agrees with HEAD on ordinary SVG icon markup [R3].

Untested areas

• Optional setting off: Producer, Setting unset and Override. These rows were not rechecked. This commit does not touch those lines.
• Open inc/views/page_header.php:229. HEAD stays in SVG mode after a breakout tag such as <p>. It also ignores MathML <mtext> and a </template> inside SVG. Malformed SVG or MathML can then hide or duplicate the fallback H1. A WordPress.org search found no plugin or theme that emits such markup.

🤖 Automated review · run code-review-agent_6adb0a2e204584.03852332.

---

🤖 Review agent — review posted ✅ on 26fded7a · changes requested · 1 finding · 21 min

Run code-review-agent_6adb0a2e204584.03852332 · trail

[pirate-bot]

🟢 Approved

The PR passes. Commit 0504889 adds the four requested SVG and MathML test rows, and it adds no new problem.

Validation details

• Files reviewed: 1/5 changed files in commit 0504889. The other four are unchanged since the last review. Unchecked row: Optional setting off: Override.
• An isolated PHP 8.3 run used the HEAD scanner. Each new row passes on HEAD and fails on its target mutant: git rev-parse HEAD; timeout 30 php /workspace/checks/L2/run.php; timeout 30 php -l /repo/tests/test-neve-hidden-title-h1… printed All four HEAD rows pass and each kills exactly its target mutant (run R1)
• The same run reports git rev-parse HEAD; timeout 30 php /workspace/checks/L2/run.php; timeout 30 php -l /repo/tests/test-neve-hidden-title-h1… printed No syntax errors detected in /repo/tests/test-neve-hidden-title-h1.php (run R1)
• A source read found no duplicate row, wrong label or stale comment.
• Checks on the unchanged files are reused from run code-review-agent_6adb0a2e204584.03852332, not rerun.

Untested areas

• Optional setting off: Override. Neve Pro or site filters that return different hide-title values per call were not checked.
• The full PHPUnit and e2e suites were not rerun. CI runs them.

🤖 Automated review · run code-review-agent_6adb1367d5ecc6.28539162.

---

🤖 Review agent — review posted ✅ on 0504889d · approved · 0 findings · 24 min

Run code-review-agent_6adb1367d5ecc6.28539162 · trail

[pirate-bot]

🟡 Changes requested

The new integration-point stack never pops on </title>, </desc> or </foreignObject>, so a later SVG H1 is missed. Fix the closing check.

Validation details

• Files reviewed: 2/5 changed files in commit 14cd6e7. The other three are unchanged since the approved review. Unchecked rows: Optional setting off: Override, Optional setting off: Producer, Optional setting off: Setting unset.
• An isolated PHP 8.3 run compared the scanner at HEAD and 0504889. timeout 30 php /workspace/checks/L1a/run.php | tee /workspace/checks/L1a/output.log printed Provider passes / 43: HEAD=43 prior=38 fix=43 (run R1)
• Headless Chrome DOMParser finds one H1 in <svg><title>t</title><style><h1>x</h1></style></svg>.
• The end-tag change matches browsers for attributes, /, newlines and end of input.

Untested areas

• Optional setting off rows (Override, Producer, Setting unset): open inc/views/page_header.php:111-126. Check that Neve Pro hide-title filters give the same answer to the buffer gate and the view. Different answers mean a missing or extra fallback H1.
• inc/views/page_header.php:44 ignores namespaces. <math><title> counts as an integration point. No producer was found.
• annotation-xml encoding="text/html" is not an integration point. No producer was found.

🤖 Automated review · run code-review-agent_6adb34ee3eaff2.27206306.

---

🤖 Review agent — review posted ✅ on 14cd6e78 · changes requested · 1 finding · 14 min

Run code-review-agent_6adb34ee3eaff2.27206306 · trail

[pirate-bot]

🟢 Approved

The PR passes. Commit 0fe4373 fixes the integration-point stack finding and adds no confirmed problem.

Validation details

• Files reviewed: 2/5 changed files in commit 0fe4373. The other three are unchanged. Unchecked rows: Optional setting off: Producer, Optional setting off: Setting unset, Optional setting off: Override.
• An isolated PHP 8.3 scanner run passed at HEAD: git -C /repo show 0fe43736b91e0a9381385b0745b1a2a889687285:tests/test-neve-hidden-title-h1.php > /workspace/checks/L1/he… printed SUMMARY HEAD: 47/47 PASS; 0 FAIL (run R1)
• The four new rows fail on 14cd6e7: git -C /repo show 0fe43736b91e0a9381385b0745b1a2a889687285:tests/test-neve-hidden-title-h1.php > /workspace/checks/L1/he… printed SUMMARY PRIOR: 43/47 PASS; 4 FAIL (run R1)

Untested areas

• At inc/views/page_header.php:240, </desc> pops the context while an unclosed <div> is open inside it. Chromium ignores that end tag, so the scan counts an H1 the page lacks. No producer of such markup was found.
• The three unchecked rows above were not rechecked.
• PHPUnit was not run. CI owns it.

🤖 Automated review · run code-review-agent_6adb3ae3c0a2c9.96967871.

---

🤖 Review agent — review posted ✅ on 0fe43736 · approved · 0 findings · 10 min

Run code-review-agent_6adb3ae3c0a2c9.96967871 · trail