Repository navigation
Conversation
Contributor
|
Are you an Automattician? Please test your changes on all WordPress.com environments to help mitigate accidental explosions.
Interested in more tips and information?
|
Contributor
|
Thank you for your PR! When contributing to Jetpack, we have a few suggestions that can help us test and review your patch:
This comment will be updated as you work on your PR and make changes. If you think that some of those checks are not needed for your PR, please explain why you think so. Thanks for cooperation 🤖 Follow this PR Review Process:
If you have questions about anything, reach out in #jetpack-developers for guidance! |
Closed
simison
force-pushed
the
simison/gutenberg-agent-skills-setup-f5043d
branch
from
November 16, 2026 17:16
350180e to
a615214
Compare
simison
force-pushed
the
simison/pnpm-version-jetpack-a47f95
branch
from
November 16, 2026 17:49
10459a6 to
12c65ae
Compare
simison
force-pushed
the
simison/gutenberg-agent-skills-setup-f5043d
branch
from
November 16, 2026 17:50
a615214 to
5c49151
Compare
simison
force-pushed
the
simison/gutenberg-agent-skills-setup-f5043d
branch
from
November 16, 2026 17:52
5c49151 to
df00d2d
Compare
…has it (#55023) The Playground importer gave every single-column primary key AUTO_INCREMENT. A plugin table keyed by text, like Novamira's `id varchar(64)`, then fails with ERROR 1063, and the whole restore fails. Read the column's EXTRA from the SQLite driver's information schema and keep AUTO_INCREMENT only where it's set.
#55012) * Settings: Point the Writing links to Features behind the settings flag With settings consolidation on, the editor's "Turn on Jetpack Blocks" link searches My Jetpack › Features for the Blocks module, and the wpcomsh notice on Settings › Writing links to Features. The search uses the module's translated name, since Features matches translated names. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ABKQhWcG89Ca835d23UqyF * Settings: Add changelog entries for the Writing links to Features Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ABKQhWcG89Ca835d23UqyF --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* My Jetpack: Link Jetpack Stats from the Google Analytics panel The Settings card this panel replaces linked "Jetpack Stats". Link it to the Stats product's manage URL, as My Jetpack's own Stats card does. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ABKQhWcG89Ca835d23UqyF * changelog Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ABKQhWcG89Ca835d23UqyF * My Jetpack: Test the Google Analytics panel's plain-text Stats fallback Without a Stats manage URL, the intro must still read as one sentence, with Jetpack Stats as plain text and no raw markup. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01ABKQhWcG89Ca835d23UqyF --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…res (#55028) SQL_Importer collected all of mysql's output with exec(). Where the client runs verbose, as on Atomic, it echoes every statement, so a large dump could exhaust PHP's memory, and errors were cut to their first 500 bytes, losing the ERROR line. Pipe the output through tail -c 4096 with mysql's exit status appended, and report mysql's ERROR line.
* fix: let Markdown block tables and code fences wrap in emails The Markdown block saves rendered HTML, so a table or a fenced code block reaches a subscription email as a plain <table> or <pre> carrying none of the block classes the email engine styles. A cell or a line holding a word too long to break then sets the width of the whole message and it scrolls sideways. Register a render_email_callback that adds word-break to table cells and white-space: pre-wrap to <pre> before handing the block to the engine's own fallback renderer, so the block keeps the wrapper and spacing every other unsupported block gets. The callback returns null when the email engine is not loaded, which leaves the engine to render the block exactly as it does today. Breaking a word only when it does not fit leaves ordinary prose alone and keeps the line breaks a code fence exists to preserve. Fixes NL-1007. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> * fix: register the Markdown block eagerly, so its email renderer is found Blocks in Jetpack_Gutenberg::$lazy_blocks have their registration PHP deferred on front-end requests and registered from pre_render_block on first render. The email editor reads render_email_callback straight off the registered block type without going through pre_render_block, so a deferred block has no callback to find when an email containing it is rendered. That rule is already written down above the list, and Jetpack_Gutenberg_Test enforces it: a lazy block must not register an e-mail renderer. Drop markdown from the list, which is the documented cost of giving it one. Markdown's registration file only registers the block, so loading it eagerly adds no work beyond the include. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * test: cover the Markdown block's registration and its absent-engine guard The coverage check flagged six uncovered lines: the body of register_block(), which no test called because the tests require the block file after init has fired, and the `return null` the guard takes when the e-mail editor is absent, which the Fallback mock makes unreachable once it is aliased in. Both are worth a test on their own. Losing render_email_callback would silently return the block to the engine's generic fallback and undo the wrapping fix, and the guard is what keeps a site running Jetpack without the e-mail editor from fataling on a class that isn't there. The guard test runs in its own process so class_exists() is genuinely false there; the mock now loads inside the one test that needs it rather than at file scope, which keeps that process clean. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix: keep a floor under narrow Markdown table cells in emails word-break lets a cell break mid-word, which is what stops a long identifier widening the email, but it also drops the cell's minimum width to a single character. Auto table layout hands width out in proportion to content, so a short column sharing a row with a long one is squeezed to that floor and its label renders one letter per line down the side of the table. Reported from a real send in iOS Gmail. Add min-width: 4em to table cells. Measured through the renderer in WebKit, which is what iOS Gmail runs: a "Code" cell goes from 29x114px to 81x42px at a 390px viewport with the table still inside the email. A code fence has no neighbouring column to be squeezed by, so <pre> keeps character-level breaking with no floor. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * chore: drop the unreachable Tag Processor guard, fix changelog case WP_HTML_Tag_Processor has shipped in core since 6.2 and the plugin declares Requires at least: 7.0, so that half of the guard could never be true where this code runs. It read as a live compatibility concern and left a branch no test could reach. The Fallback check stays: that class comes from a separate package whose presence really can vary. The changelog entry now capitalises after its component prefix, matching its siblings and what AGENTS.md asks for. Both raised by the a8c-review CLI. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix: lower the Markdown table cell floor so short columns still fit The 4em floor applied to every cell, so a table of short columns that fitted before it existed no longer did. Measured through the renderer in WebKit at a 390px viewport, a five column table of one-character values went from 342px to 454px, scrolling sideways — the symptom this branch set out to remove. 3em is the largest floor that leaves five columns inside the content width while still keeping a short label readable beside a long unbreakable value. Matches the same correction in the email editor package, raised by CodeRabbit on woocommerce/woocommerce#70892. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * test: measure the cell floor against the width a phone gives the table The regression assertion compared against the 612px desktop content width, which the 4em floor it was written to reject clears comfortably: 5 columns at 4em comes to 400px. The test passed with the exact value that caused the reported overflow, so it guarded nothing. Measure against 342px instead, the width left after the email's root padding on a 390px phone, and count the collapsed borders. A second assertion pins that threshold so the limit can't quietly stop biting. Matches the same correction in the email editor package, raised by CodeRabbit on woocommerce/woocommerce#70892. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix: give Markdown tables a fixed layout instead of a per-cell minimum A per-cell minimum width adds up across the row, so a table of eight short columns rendered 400px inside the 342px a 390px phone leaves the email and scrolled sideways. Fixed layout and a full width divide that space between the columns instead, which holds at any column count and still leaves a short label beside a long value a column of its own. Measured in WebKit at a 390px viewport inside a 342px container, for two columns short+long and long+short, five, six and eight short columns, five short plus one long, short/long/short, and a code fence. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> * fix: give Markdown tables the width attribute Outlook reads Outlook's Word engine may ignore a table's CSS width, so the email engine's own table renderer also sets width="100%". Markdown tables skip that renderer, so set it here too, keeping any width the author gave. Also say in the changelog that tables now span the full width with equal columns, since that is what a reader will notice, and trim the comments around the email styles to the reason each rule is there. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> * fix: always give Markdown tables a full-width attribute The inline style already forces the table to 100%, so keeping an author's own width attribute only made Outlook disagree with every other client. Also drop a test assertion that repeated the table layout check. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
…53562) WordPress.com answers stats/file-downloads with a 404 (invalid_blog) for every Jetpack-connected site, self-hosted or WordPress.com Business, so get-top-content with type=downloads failed there with the generic "Failed to fetch Stats from WPCOM". It now returns jetpack_stats_invalid_type with a clear message on those sites, without calling WordPress.com, and its description says downloads only work on WordPress.com sites that do not use Jetpack. Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
…4782) * SEO: Clarify site-wide indexing restrictions and item preferences * SEO: Link indexing notices to the available visibility control --------- Co-authored-by: Chris Jean <chris.jean@automattic.com>
…54973) * SEO: Avoid duplicate dashboard store registration and config errors * SEO: Describe dashboard fixes in user-facing changelogs --------- Co-authored-by: Chris Jean <chris.jean@automattic.com>
…rn the Agent on (#55033) The notice's Enable link leads to the WordPress.com AI tools settings, which the dashboard answers with a 404 for anyone who cannot manage the site: non-admins, super admins who are not members of the blog, and P2 and VIP sites. Compute that in PHP as a new ai-sidebar-agent-enable extension flag, and keep the legacy Improve with AI panel for people who lack it while the Agent is off.
…54972) * SEO: Detect static files at the public root of subdirectory installs * SEO: Normalize filesystem paths before resolving the public root * SEO: Align static-file guidance with the public directory --------- Co-authored-by: Chris Jean <chris.jean@automattic.com>
* Block Delimiter: Add Block_Scanner::remove_blocks() to remove blocks of a type Removes every block of one type from a serialized block document, including nested and void blocks. Nothing calls it yet, so sites that load an older copy of the package are unaffected. * Block Delimiter: Narrow the remove_blocks() docblock on line break cleanup
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Depends on pnpm v12 #55013
Depends on upstream agent skills added in WordPress/gutenberg#86737
Experimental; not for review. I'm developing method to include agent skills and testing how it works with monorepo like Jetpack.
Proposed changes
pnpm install.Related product discussion/links
Does this pull request change what data or activity we track or use?
Testing instructions
.claude/skills/and.agents/skillsfolders afterpnpm install