[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6991DABP5586SNSFX3PWNEQ":3,"skill-same:01M6991DABP5586SNSFX3PWNEQ":71},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":39,"linkOnly":40,"markdownContent":41,"markdownBody":42,"frontMatter":43,"refName":47,"commitSha":48,"skillDirectory":49,"forksCount":50,"repositoryHtmlUrl":51,"repositoryHomepageUrl":52,"ownerHtmlUrl":53,"ownerBlogUrl":54,"ownerDescription":55,"updatedAt":56,"install":57,"resources":70},"01M6991DABP5586SNSFX3PWNEQ","write-implementation-brief",null,"Write the Implementation Brief and Test Coverage sections of a Site Kit issue from its acceptance criteria. Use when the user asks to write, draft, adjust, update, or fill in an implementation brief and\u002For test coverage for an issue (e.g. \"write the implementation brief for #12345\", \"fill in the brief and test coverage for this issue\") in the google\u002Fsite-kit-wp repo.","google\u002Fsite-kit-wp","google","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1342004?v=4","01M698VYEFDM0VJBZDYYJ2T83S","Google","01M69913CJ7EWER6YJA1HXMZQV",1405,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002Fdevelop\u002F.agents\u002Fskills\u002Fwrite-implementation-brief\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27,28,29,30,31,32,9,33,34,35,36,37,38],"write implementation brief","brief","write","implementation","use","when","fill","for","number","kit","adjust","draft","acceptance criteria","test coverage","coverage","repo","Bash","acceptance","test","site kit","2026-11-10 10:12:30",false,"---\nname: write-implementation-brief\ndescription: >\n  Write the Implementation Brief and Test Coverage sections of a Site Kit issue from its\n  acceptance criteria. Use when the user asks to write, draft, adjust, update, or fill in an\n  implementation brief and\u002For test coverage for an issue (e.g. \"write the implementation brief\n  for #12345\", \"fill in the brief and test coverage for this issue\") in the google\u002Fsite-kit-wp repo.\nargument-hint: \"[issue-number-or-path]\"\nallowed-tools: Bash, Read, Grep, Glob, Edit, Write\n---\n\n# Write an implementation brief\n\nWrite the **Implementation Brief** and **Test Coverage** sections for **$ARGUMENTS** — either a\nGitHub issue number, or a path the user supplied to a local issue file.\n\nThe brief you write is the spec that the `implement-issue` skill executes and the `review-pr`\nskill grades against. Write it so an engineer who has never read the design doc can implement\nthe issue from the brief plus the codebase alone.\n\nThe brief gives **technical direction, not step-by-step instructions**. Keep it short, but never\nso short that the scope of the work is hidden.\n\n---\n\n## Step 1 — Read the issue\n\nThe issue comes to you one of two ways, and only the user can tell you which:\n\n- **An issue number** — fetch it with `gh issue view \u003Cnumber> --json title,body,labels`.\n- **A path to a local issue file** — read that path. Do not go looking for it: such a file has no\n  fixed home, so the user must give you the path.\n\nIf the user gave you **neither** a number nor a path, ask for one before doing anything else.\n\nParse whichever you get using the section map in `.agents\u002Fskills\u002Fimplement-issue\u002FSKILL.md` Step 1.\n\nYou are writing exactly two sections:\n\n- **Implementation Brief** — between `## Implementation Brief` and `### Test Coverage`.\n- **Test Coverage** — between `### Test Coverage` and `## QA Brief`.\n\nThe **Acceptance criteria** are the contract, and they describe observable behavior only: what a\ntester can see by using the feature. The mechanism that produces that behavior is the brief's to\nspecify, so expect to add detail the criteria do not carry. Everything in the brief must serve a\ncriterion, and the brief may not add behavior a user or a tester would notice that no criterion\nasks for.\n\n**Stop and ask the user** if the acceptance criteria are missing, ambiguous, or contradict each\nother. A brief written over guessed criteria is worse than no brief. Writing the criteria is a\nseparate job with its own skill, `write-issue-requirements` — offer it rather than filling the gap here.\n\nFeature Description and Acceptance criteria sit under a \"managed by moderators only\" notice in\nthe issue template — do not edit them. QA Brief and Changelog entry are filled in later by other\nroles; leave them alone unless asked.\n\n## Step 2 — Read the source material\n\n- **The design doc.** It carries the rationale, the rejected alternatives, and the constraints\n  that an acceptance criterion compresses into a single line, so read it whenever one exists.\n  Use it when the issue links one or the user gives you a path; otherwise **ask the user for\n  it** — if you already have to ask for the issue itself (Step 1), ask for both at once. Plenty\n  of issues have no design doc: when the answer is \"there isn't one\", say so and write the brief\n  from the acceptance criteria and the code. Never stall on a design doc, and never invent a\n  path to one.\n- **Sibling issues in the same epic** — they tell you where this issue's scope stops and the\n  next one begins. Ask the user for their numbers so you can cross-reference them (see\n  *Referencing another issue*, below).\n\n## Step 3 — Verify everything against the code\n\nDo not write a brief from the issue text alone. Open every file the brief will touch and\nconfirm, by reading it:\n\n- Every class, method, constant, hook, filter, option and script handle you name exists and is\n  spelled exactly right.\n- The pattern you tell someone to follow (e.g. \"mirror `X::register_script()`\") actually looks\n  the way you describe.\n- Hook ordering and firing conditions, whenever the brief depends on them — trace the hook to\n  its `do_action()` \u002F `apply_filters()` call site rather than trusting the design doc.\n- Which existing tests, fixtures, snapshots or inline data the change will move.\n- **That the build can compile the kind of file you are about to name.** Open the build config\n  that will include the new file, plus the lint config and `tsconfig.json`. A bundle can define\n  its own loader rules instead of using the shared `createRules()`, and where those rules do not\n  already match the extension you are naming, adding a rule is part of the brief. Check the\n  globals the new file reads in the same pass: a global with no declaration in\n  `assets\u002Fjs\u002Ftypes\u002Fglobals.d.ts` fails `npm run typecheck` until one is added.\n\nWhere the code contradicts the design doc, the code wins: write the brief against reality and\nreport the discrepancy (Step 7).\n\n## Step 4 — Load the relevant convention docs\n\nRead **only** what the issue touches, using the scope map in `.agents\u002Fskills\u002Fimplement-issue\u002FSKILL.md` Step 3. The\nbrief should not restate those conventions — it should be written so that following it produces\nconforming code.\n\n## Referencing another issue — always ask for the number\n\nEvery reference to another issue is a real GitHub number (`#12345`) — prerequisites, siblings in\nthe same epic, the extension point this brief leaves open, the follow-up that lands the rest.\nThere is no acceptable indirect form: not \"issue 5\" from a design doc's ordering, not \"the next\nissue\", not \"the issue that adds the datastore\".\n\n**Ask the user for the number.** Never infer it from a design doc's ordering, never guess it from\na nearby issue number, and never leave a placeholder to be filled in later:\n\n- **The issue is already in GitHub** — ask the user for its number, then confirm it is the issue\n  you mean with `gh issue view \u003Cnumber> --json title` before citing it.\n- **The issue only exists locally** — a markdown file that has not been filed yet. Say so, and\n  ask the user to create it in GitHub and give you the new number. Wait for it: a cross-reference\n  is the one thing you cannot write around.\n- **The user has no number to give** — drop the cross-reference and name the deliverable instead\n  (``the store introduced with `getAudienceSettings` ``). Report the dropped reference in Step 7.\n\nWork out every issue you will need to cite before you start writing, and ask for all of the\nnumbers in one message — together with the issue and design doc you already have to ask for\n(Steps 1 and 2) — rather than stopping repeatedly.\n\n## Step 5 — Write the Implementation Brief\n\n### Structure\n\n```\n\u003Cone or two sentences of context every checkbox shares, e.g. the action they all call>\n\n* [ ] \u003Cone change — it may cover several files>\n* [ ] \u003Cone change>\n```\n\n- One checkbox per change, not per file: \"Move `FeedbackMenu` and `ThumbsSurveyTrigger` to a new\n  `assets\u002Fjs\u002Fcomponents\u002Ffeedback\u002F` directory, with their tests and stories\".\n- Give a path only where the reader needs it to find the place. Otherwise use the name\n  (`ThumbsFeedback`).\n- Order the checkboxes the way someone would work through them: shared\u002Fbase changes, then new\n  classes, then registration, then build config, then components, with SCSS last.\n\nNest a second level when one bullet covers several related behaviors, ending the parent in a\ncolon:\n\n```\n* [ ] Submit behavior:\n  * \u003Con click>\n  * \u003Con success>\n  * \u003Con failure>\n```\n\n### Content\n\n**Give direction, not steps.** Write only what an engineer who knows Site Kit could not work out\nfrom the code and the acceptance criteria: the decisions (where code lives, what to reuse or\nreplace, new signatures and data shapes) and the places that are not obvious, such as a file the\nacceptance criteria imply but do not name. Leave out:\n\n- follow-on work after a move or rename — imports, usages, tests, stories, story titles;\n- a value the acceptance criteria define — point at the criterion instead;\n- names that no code outside this issue depends on — say what to add and where (\"add constants\n  for the feedback namespaces and keys next to the existing vote ID constants in: …\"), and let\n  the engineer name them.\n\n**Keep the scope visible.** When one checkbox covers many places, list them, so the reader can\nsize the work. Cut the *how* at each place, never the *where*:\n\n- Too short: \"Pass the namespace and group wherever `FeedbackPrompt` or `ThumbsFeedback` is used.\"\n- Right: \"…is used: `FeatureRelevancyFeedback`, `OnlineStorePerformanceWidget`,\n  `LeadGenerationPerformanceWidget`, `PrimaryActionRow` and the Recent activity `InsightNotice`.\"\n\n**Write in plain, simple words — this matters more than anything else in this section.** The\nbrief is read by people: the engineer who implements it, the reviewer who grades the PR against\nit, and a moderator who has never read the design doc. Many of them do not speak English as a\nfirst language. Every sentence must be clear on the first reading:\n\n- **Use the simplest word that is still accurate.** Short sentences. One idea per sentence. No\n  decoration, no \"notably\", \"crucially\", \"simply\", \"of course\", \"elegant\", \"robust\".\n- **No idioms, metaphors, or cultural references.** Not \"standard furniture\", \"dead weight\",\n  \"burns the flag\", \"under the hood\", \"out of the box\", \"for free\", \"a natural no-op\". Write what\n  actually happens instead: \"the listener finds no matching elements, so it does nothing\".\n- **Explain a technical term the first time you use it, or do not use it.** \"the value lands in\n  an opaque `pathname`\" says nothing to the reader. \"`hostname` is empty and the whole recipient\n  stays in `pathname` as one unsplit string with no leading slash — `tel:+15551234567` gives\n  `pathname` `+15551234567`\" says all of it.\n- **Finish every sentence.** \"`https:\u002F\u002Fnotwa.me\u002F1555` must not classify\" leaves out *as what*.\n  Write \"…is not a contact link, so nothing is emitted\".\n- **Name who does what to what.** \"the classifier short-circuits `outbound_link_click`\" can be\n  read in two opposite directions. \"when a link is classified as a contact link, the listener\n  emits `contact_link_click` and returns, so the outbound handler never receives that anchor\"\n  can be read in only one.\n- **List the members of any set you name.** If you write \"an allowlist\", \"the excluded hosts\",\n  \"the usual guards\", write out what is in it, inline. A set the reader has to guess at is the\n  same as no set at all.\n\nDelete every bullet the engineer would do anyway, and check that every area the change touches\nis named.\n\nRead each bullet once more before you finish, and rewrite every sentence that needs a second\nreading. This applies to your reply to the user (Step 7) as much as to the issue text.\n\n**Say what to do, and nothing else.** Every bullet is an instruction. Leave a genuine\nimplementation choice open rather than inventing a detail, but never leave the implementer to\nsearch for something the codebase has already decided.\n\n**Name the real thing.** Wherever a symbol already exists, name it exactly:\n\n- classes, methods, hooks, filters, constants, script handles, datapoint names (`GET:ctas`);\n- datastore selectors and actions (`getCTAs()`), and the form\u002Fstore constants they use;\n- components to reuse (`SpinnerButton`, `ProgressBar`, `Notice`), their props (`helperText`),\n  and CSS class names (`mdc-text-field--error`);\n- inline-data globals and keys (`wpPrivacyURL` on `_googlesitekitBaseData`).\n\nGive a signature only when the brief adds or changes it.\n\n**Spell out every name you invent.** Name a new thing only when other code depends on the name\n(see *Give direction, not steps*). Write each name you give in full words — no abbreviations,\nno truncations, no initials. `readThresholdPct` and `readMinSeconds` each cost the reader a guess:\nwhat does `Pct` stand for, and is `Min` \"minutes\" or \"minimum\"? `readTimeThresholdPercent` and\n`minimumReadTimeSeconds` cost nothing and can only be read one way. The short forms that stay are\nthe ones already used across the codebase (`ID`, `URL`, `HTML`) and any existing name, which you\ncopy exactly as it is. See `docs\u002Fcontext\u002Fphp\u002Fnaming-conventions.md`.\n\n**Point at an exemplar instead of describing one.** \"following the pattern in\n`includes\u002FModules\u002FAnalytics_4\u002FDatapoints\u002F`\", \"following `publications.js`\", \"following how\n`Analytics_4` registers `POST:create-audience`\". One reference beats a paragraph of description.\n\n**Cross-reference sibling issues inline** wherever the work meets theirs — `(added in #12950)`,\n`(see #13005)`, \"the create variant is #13022\". Use it for prerequisites the brief depends on\nand for extension points it deliberately leaves open. Every number comes from the user, never\nfrom your own inference — see *Referencing another issue*, above.\n\n**Fence the scope in one line** when an adjacent concern could reasonably be pulled in: \"No\nexpress-flow routing\u002Fgating changes in this step.\" \"The live API wiring lands in the API\nintegration work.\" \"Keep shared stepper\u002Flayout styles untouched.\"\n\n**For UI work**, reference the design (a Figma node link, or \"per the design\") rather than\nrestating measurements, and spell out the interaction states: what disables the CTA, the pending\nstate, what happens on success, what renders on error. Name the SCSS partial to create or extend\nand the index file to import it into.\n\nDo **not** include:\n\n- **Rationale.** No \"because\", no \"this keeps X out of Y\", no trade-off discussion, no\n  restatement of a decision's benefits. The design doc and the acceptance criteria hold the\n  *why*; the brief holds the *what*. A reason is the exception, allowed in two places only:\n  where the brief settles a real technical choice an engineer would otherwise stop and\n  question — keep it to a clause — and for a section the issue template asks for that needs\n  nothing (see the next bullet).\n- **Work that isn't needed.** No \"no migration needed\", \"no new settings\", \"no REST route\" for a\n  concern the issue never raised. Silence already says it. The exception is what the template\n  asks for by name — automated tests, and Storybook and VRT changes: when those aren't needed,\n  say so and say why (Step 6).\n- **Routine commands.** No lint, build, test or VRT invocations — the `implement-issue` skill's\n  Step 6 owns verification, and every engineer already knows to build. Mention a command only when it\n  is unusual and specific to this issue.\n- **Links to local design docs.** They don't resolve for someone reading the issue on GitHub;\n  restate the constraint in one clause instead. External links — Figma, a hosted design doc —\n  are fine.\n- **Plain-text line numbers.** No `includes\u002FCore\u002FModules\u002FModules.php:212`, no \"see line 212\", no\n  `file:line` anywhere in the two sections. The number is wrong as soon as anyone edits the file,\n  and a reader on GitHub cannot click it. When a bullet has to point at an exact place in the\n  code, use a GitHub permalink instead — a URL pinned to a commit SHA, with the line or range on\n  the end:\n  `https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002F\u003Ccommit SHA>\u002Fincludes\u002FCore\u002FModules\u002FModules.php#L212-L218`.\n  Get one by opening the file on GitHub and pressing `y`, or with\n  `gh browse --no-browser --commit=\u003Ccommit SHA> \u003Cpath>` and adding the `#L\u003Cstart>-L\u003Cend>` yourself.\n  A URL built on a branch name (`\u002Fblob\u002Fmain\u002F`, `\u002Fblob\u002Fdevelop\u002F`) is **not** a permalink — it points\n  at different lines next week — so never use one. Naming the symbol is still better than linking\n  to it: link only when there is no name to give, such as an unnamed block inside a long function.\n- **Restated acceptance criteria**, a recap of the Feature Description, or background on how the\n  existing system works. This covers the user-facing copy the criteria define — headings,\n  descriptions, CTA labels, error text. Copy gets reworded during brief review, and a verbatim\n  quote in the brief goes stale while the criterion it came from stays current; point at the\n  criterion instead (\"the empty-state heading from the acceptance criteria\"). A string that\n  already exists in the code is not an acceptance criterion — name it exactly, as above.\n- Prose paragraphs.\n\nDo include, when they apply:\n\n- **The right extension on every new frontend file.** New files on the JavaScript side are\n  written in TypeScript: `.ts`, or `.tsx` when the file returns JSX, with co-located tests as\n  `.test.ts` \u002F `.test.tsx`. Never ask for a new `.js` or `.jsx` file. Name the types the new file\n  exports the same way you name any other symbol — the union of allowed values, the interface a\n  table row must match, the function signature with its return type\n  (`classifyContactLink( anchor: HTMLAnchorElement ): ContactLinkType | null`). An existing `.js`\n  file that the issue only edits stays as it is, unless the issue asks for a rename.\n- The concrete shape of any data the change publishes — array keys, JSON payload, inline-script\n  global, selector signature.\n- Changes to existing files that the acceptance criteria imply but do not name.\n\n## Step 6 — Write the Test Coverage\n\nKeep it short — a handful of bullets, not a test plan. Two shapes, both in use:\n\n```\n* Add tests for `\u003Csource file>` covering:\n  * \u003Ccase>\n  * \u003Ccase>\n\n* JS tests in `\u003Cfile>.test.ts` covering the `\u003Cname>` action and the `\u003Cname>` selector.\n\n* Update the tests for each component changed above, so that they check \u003Cbehavior>.\n```\n\n- One bullet per test file or area, with the cases nested under it. Name the test file when it\n  exists or its name follows from the source: co-located next to the source — `*.test.ts` \u002F\n  `*.test.tsx` for every new test file, `*.test.js` only when you are adding cases to a JavaScript\n  test file that already exists — and `*Test.php` under the mirroring path in\n  `tests\u002Fphpunit\u002Fintegration\u002F`. When the same check applies to every file, write one bullet for\n  all of them.\n- Phrase each case as the behavior or outcome, not the mechanics — \"the privacy policy field\n  falls back to the WordPress privacy policy URL when the publication value is missing\", not\n  \"mock the store and assert the input value\".\n- Cover every acceptance criterion at least once, including the negative cases: precondition\n  unmet, feature disabled, invalid input, empty or error response.\n- **Storybook stories belong here, not in the brief.** \"Add a Story for `\u003CComponent>`\", or list\n  the states when there is more than one: \"(default, loading, error)\".\n- **VRT belongs here too, not in a section of its own.** Say what the change does to the VRT\n  scenarios and the reference images — a new scenario for a new story, new reference images for\n  a component whose rendering changes. \"No VRT changes expected\" is a complete answer when nothing\n  the VRT job renders can move. Do not give screenshot paths.\n- **Say so when nothing is needed, and say why.** Never leave Test Coverage empty, and never drop\n  the Storybook or VRT line in silence. When the change needs no tests or no story, write one\n  bullet that says so with the reason — \"No tests required because the change only edits\n  documentation\", \"No Storybook story required because the change adds no UI\". VRT is the\n  exception: a bare \"No VRT changes expected\" needs no reason. An empty section reads as a\n  forgotten section, and a moderator cannot tell the two apart.\n- **Never write `N\u002FA` for tests.** Tests are almost never inapplicable, so `N\u002FA` reads as a\n  skipped question rather than an answer — give the reason instead, as above. The VRT line is\n  where `N\u002FA` is allowed, because there it is often literally true: PHP-only changes, tooling and\n  build config, anything the VRT job never renders.\n- Name any new fixture, fake or mock the coverage needs, and the directory it belongs in.\n- Call out **existing** tests the change will break, so the work is budgeted rather than\n  discovered mid-implementation — \"Fix any failing tests\u002Fsnapshots\" when the blast radius isn't\n  knowable up front.\n- No commands.\n\n## Step 7 — Report what did not go in the brief\n\nEverything you cut for brevity, and everything you learned while verifying, belongs in your\nreply to the user — not in the issue:\n\n- Discrepancies between the design doc and the code, with the `file:line` you checked.\n- Consequences the acceptance criteria did not anticipate: a metric that will shift, a default\n  that changes for every install, an existing test that will start failing.\n- Anything you left as an open decision, and the assumption the brief is written under.\n- Every issue number you were given and what you cited it for, any cross-reference you dropped\n  because no number was available, and any locally-written issue that still has to be filed\n  before the brief's text is final.\n\n---\n\n## Guardrails\n\n- **Write only the two sections.** Leave Feature Description, Acceptance criteria, QA Brief and\n  Changelog entry untouched.\n- **Direction, not steps.** Leave out what the engineer would do anyway; name every area the\n  change touches — see Step 5.\n- **Plain, simple words.** No idioms, no metaphors, no unexplained jargon, no half sentences.\n  Name who does what to what, and list the members of every set you name — see Step 5.\n- **New frontend files are TypeScript.** `.ts` \u002F `.tsx` sources, `.test.ts` \u002F `.test.tsx` tests,\n  and a build-config change whenever the bundle cannot compile them yet.\n- **Names are full words.** Name a new thing only when other code depends on it, in full words\n  — no `Pct`, no `Min`, no initials — see Step 5.\n- **Don't publish.** Do not edit the GitHub issue or post a comment unless the user explicitly\n  asks. Produce the text — or update the local issue file you were given — and let them place it.\n- **Don't expand scope.** If a change looks necessary but no acceptance criterion covers it,\n  raise it in Step 7 instead of adding it silently.\n- **No indirect issue references.** Every cross-reference is a real GitHub number you asked the\n  user for, never a design-doc position, a relative pointer or a placeholder.\n- **No plain-text line numbers.** Every pointer into the code is a GitHub permalink pinned to a\n  commit SHA, never `path\u002Fto\u002FFile.php:212` and never a `\u002Fblob\u002F\u003Cbranch>\u002F` URL — see Step 5.\n- **Verify before naming.** Never reference a symbol, path or hook you have not opened.\n","# Write an implementation brief\n\nWrite the **Implementation Brief** and **Test Coverage** sections for **$ARGUMENTS** — either a\nGitHub issue number, or a path the user supplied to a local issue file.\n\nThe brief you write is the spec that the `implement-issue` skill executes and the `review-pr`\nskill grades against. Write it so an engineer who has never read the design doc can implement\nthe issue from the brief plus the codebase alone.\n\nThe brief gives **technical direction, not step-by-step instructions**. Keep it short, but never\nso short that the scope of the work is hidden.\n\n---\n\n## Step 1 — Read the issue\n\nThe issue comes to you one of two ways, and only the user can tell you which:\n\n- **An issue number** — fetch it with `gh issue view \u003Cnumber> --json title,body,labels`.\n- **A path to a local issue file** — read that path. Do not go looking for it: such a file has no\n  fixed home, so the user must give you the path.\n\nIf the user gave you **neither** a number nor a path, ask for one before doing anything else.\n\nParse whichever you get using the section map in `.agents\u002Fskills\u002Fimplement-issue\u002FSKILL.md` Step 1.\n\nYou are writing exactly two sections:\n\n- **Implementation Brief** — between `## Implementation Brief` and `### Test Coverage`.\n- **Test Coverage** — between `### Test Coverage` and `## QA Brief`.\n\nThe **Acceptance criteria** are the contract, and they describe observable behavior only: what a\ntester can see by using the feature. The mechanism that produces that behavior is the brief's to\nspecify, so expect to add detail the criteria do not carry. Everything in the brief must serve a\ncriterion, and the brief may not add behavior a user or a tester would notice that no criterion\nasks for.\n\n**Stop and ask the user** if the acceptance criteria are missing, ambiguous, or contradict each\nother. A brief written over guessed criteria is worse than no brief. Writing the criteria is a\nseparate job with its own skill, `write-issue-requirements` — offer it rather than filling the gap here.\n\nFeature Description and Acceptance criteria sit under a \"managed by moderators only\" notice in\nthe issue template — do not edit them. QA Brief and Changelog entry are filled in later by other\nroles; leave them alone unless asked.\n\n## Step 2 — Read the source material\n\n- **The design doc.** It carries the rationale, the rejected alternatives, and the constraints\n  that an acceptance criterion compresses into a single line, so read it whenever one exists.\n  Use it when the issue links one or the user gives you a path; otherwise **ask the user for\n  it** — if you already have to ask for the issue itself (Step 1), ask for both at once. Plenty\n  of issues have no design doc: when the answer is \"there isn't one\", say so and write the brief\n  from the acceptance criteria and the code. Never stall on a design doc, and never invent a\n  path to one.\n- **Sibling issues in the same epic** — they tell you where this issue's scope stops and the\n  next one begins. Ask the user for their numbers so you can cross-reference them (see\n  *Referencing another issue*, below).\n\n## Step 3 — Verify everything against the code\n\nDo not write a brief from the issue text alone. Open every file the brief will touch and\nconfirm, by reading it:\n\n- Every class, method, constant, hook, filter, option and script handle you name exists and is\n  spelled exactly right.\n- The pattern you tell someone to follow (e.g. \"mirror `X::register_script()`\") actually looks\n  the way you describe.\n- Hook ordering and firing conditions, whenever the brief depends on them — trace the hook to\n  its `do_action()` \u002F `apply_filters()` call site rather than trusting the design doc.\n- Which existing tests, fixtures, snapshots or inline data the change will move.\n- **That the build can compile the kind of file you are about to name.** Open the build config\n  that will include the new file, plus the lint config and `tsconfig.json`. A bundle can define\n  its own loader rules instead of using the shared `createRules()`, and where those rules do not\n  already match the extension you are naming, adding a rule is part of the brief. Check the\n  globals the new file reads in the same pass: a global with no declaration in\n  `assets\u002Fjs\u002Ftypes\u002Fglobals.d.ts` fails `npm run typecheck` until one is added.\n\nWhere the code contradicts the design doc, the code wins: write the brief against reality and\nreport the discrepancy (Step 7).\n\n## Step 4 — Load the relevant convention docs\n\nRead **only** what the issue touches, using the scope map in `.agents\u002Fskills\u002Fimplement-issue\u002FSKILL.md` Step 3. The\nbrief should not restate those conventions — it should be written so that following it produces\nconforming code.\n\n## Referencing another issue — always ask for the number\n\nEvery reference to another issue is a real GitHub number (`#12345`) — prerequisites, siblings in\nthe same epic, the extension point this brief leaves open, the follow-up that lands the rest.\nThere is no acceptable indirect form: not \"issue 5\" from a design doc's ordering, not \"the next\nissue\", not \"the issue that adds the datastore\".\n\n**Ask the user for the number.** Never infer it from a design doc's ordering, never guess it from\na nearby issue number, and never leave a placeholder to be filled in later:\n\n- **The issue is already in GitHub** — ask the user for its number, then confirm it is the issue\n  you mean with `gh issue view \u003Cnumber> --json title` before citing it.\n- **The issue only exists locally** — a markdown file that has not been filed yet. Say so, and\n  ask the user to create it in GitHub and give you the new number. Wait for it: a cross-reference\n  is the one thing you cannot write around.\n- **The user has no number to give** — drop the cross-reference and name the deliverable instead\n  (``the store introduced with `getAudienceSettings` ``). Report the dropped reference in Step 7.\n\nWork out every issue you will need to cite before you start writing, and ask for all of the\nnumbers in one message — together with the issue and design doc you already have to ask for\n(Steps 1 and 2) — rather than stopping repeatedly.\n\n## Step 5 — Write the Implementation Brief\n\n### Structure\n\n```\n\u003Cone or two sentences of context every checkbox shares, e.g. the action they all call>\n\n* [ ] \u003Cone change — it may cover several files>\n* [ ] \u003Cone change>\n```\n\n- One checkbox per change, not per file: \"Move `FeedbackMenu` and `ThumbsSurveyTrigger` to a new\n  `assets\u002Fjs\u002Fcomponents\u002Ffeedback\u002F` directory, with their tests and stories\".\n- Give a path only where the reader needs it to find the place. Otherwise use the name\n  (`ThumbsFeedback`).\n- Order the checkboxes the way someone would work through them: shared\u002Fbase changes, then new\n  classes, then registration, then build config, then components, with SCSS last.\n\nNest a second level when one bullet covers several related behaviors, ending the parent in a\ncolon:\n\n```\n* [ ] Submit behavior:\n  * \u003Con click>\n  * \u003Con success>\n  * \u003Con failure>\n```\n\n### Content\n\n**Give direction, not steps.** Write only what an engineer who knows Site Kit could not work out\nfrom the code and the acceptance criteria: the decisions (where code lives, what to reuse or\nreplace, new signatures and data shapes) and the places that are not obvious, such as a file the\nacceptance criteria imply but do not name. Leave out:\n\n- follow-on work after a move or rename — imports, usages, tests, stories, story titles;\n- a value the acceptance criteria define — point at the criterion instead;\n- names that no code outside this issue depends on — say what to add and where (\"add constants\n  for the feedback namespaces and keys next to the existing vote ID constants in: …\"), and let\n  the engineer name them.\n\n**Keep the scope visible.** When one checkbox covers many places, list them, so the reader can\nsize the work. Cut the *how* at each place, never the *where*:\n\n- Too short: \"Pass the namespace and group wherever `FeedbackPrompt` or `ThumbsFeedback` is used.\"\n- Right: \"…is used: `FeatureRelevancyFeedback`, `OnlineStorePerformanceWidget`,\n  `LeadGenerationPerformanceWidget`, `PrimaryActionRow` and the Recent activity `InsightNotice`.\"\n\n**Write in plain, simple words — this matters more than anything else in this section.** The\nbrief is read by people: the engineer who implements it, the reviewer who grades the PR against\nit, and a moderator who has never read the design doc. Many of them do not speak English as a\nfirst language. Every sentence must be clear on the first reading:\n\n- **Use the simplest word that is still accurate.** Short sentences. One idea per sentence. No\n  decoration, no \"notably\", \"crucially\", \"simply\", \"of course\", \"elegant\", \"robust\".\n- **No idioms, metaphors, or cultural references.** Not \"standard furniture\", \"dead weight\",\n  \"burns the flag\", \"under the hood\", \"out of the box\", \"for free\", \"a natural no-op\". Write what\n  actually happens instead: \"the listener finds no matching elements, so it does nothing\".\n- **Explain a technical term the first time you use it, or do not use it.** \"the value lands in\n  an opaque `pathname`\" says nothing to the reader. \"`hostname` is empty and the whole recipient\n  stays in `pathname` as one unsplit string with no leading slash — `tel:+15551234567` gives\n  `pathname` `+15551234567`\" says all of it.\n- **Finish every sentence.** \"`https:\u002F\u002Fnotwa.me\u002F1555` must not classify\" leaves out *as what*.\n  Write \"…is not a contact link, so nothing is emitted\".\n- **Name who does what to what.** \"the classifier short-circuits `outbound_link_click`\" can be\n  read in two opposite directions. \"when a link is classified as a contact link, the listener\n  emits `contact_link_click` and returns, so the outbound handler never receives that anchor\"\n  can be read in only one.\n- **List the members of any set you name.** If you write \"an allowlist\", \"the excluded hosts\",\n  \"the usual guards\", write out what is in it, inline. A set the reader has to guess at is the\n  same as no set at all.\n\nDelete every bullet the engineer would do anyway, and check that every area the change touches\nis named.\n\nRead each bullet once more before you finish, and rewrite every sentence that needs a second\nreading. This applies to your reply to the user (Step 7) as much as to the issue text.\n\n**Say what to do, and nothing else.** Every bullet is an instruction. Leave a genuine\nimplementation choice open rather than inventing a detail, but never leave the implementer to\nsearch for something the codebase has already decided.\n\n**Name the real thing.** Wherever a symbol already exists, name it exactly:\n\n- classes, methods, hooks, filters, constants, script handles, datapoint names (`GET:ctas`);\n- datastore selectors and actions (`getCTAs()`), and the form\u002Fstore constants they use;\n- components to reuse (`SpinnerButton`, `ProgressBar`, `Notice`), their props (`helperText`),\n  and CSS class names (`mdc-text-field--error`);\n- inline-data globals and keys (`wpPrivacyURL` on `_googlesitekitBaseData`).\n\nGive a signature only when the brief adds or changes it.\n\n**Spell out every name you invent.** Name a new thing only when other code depends on the name\n(see *Give direction, not steps*). Write each name you give in full words — no abbreviations,\nno truncations, no initials. `readThresholdPct` and `readMinSeconds` each cost the reader a guess:\nwhat does `Pct` stand for, and is `Min` \"minutes\" or \"minimum\"? `readTimeThresholdPercent` and\n`minimumReadTimeSeconds` cost nothing and can only be read one way. The short forms that stay are\nthe ones already used across the codebase (`ID`, `URL`, `HTML`) and any existing name, which you\ncopy exactly as it is. See `docs\u002Fcontext\u002Fphp\u002Fnaming-conventions.md`.\n\n**Point at an exemplar instead of describing one.** \"following the pattern in\n`includes\u002FModules\u002FAnalytics_4\u002FDatapoints\u002F`\", \"following `publications.js`\", \"following how\n`Analytics_4` registers `POST:create-audience`\". One reference beats a paragraph of description.\n\n**Cross-reference sibling issues inline** wherever the work meets theirs — `(added in #12950)`,\n`(see #13005)`, \"the create variant is #13022\". Use it for prerequisites the brief depends on\nand for extension points it deliberately leaves open. Every number comes from the user, never\nfrom your own inference — see *Referencing another issue*, above.\n\n**Fence the scope in one line** when an adjacent concern could reasonably be pulled in: \"No\nexpress-flow routing\u002Fgating changes in this step.\" \"The live API wiring lands in the API\nintegration work.\" \"Keep shared stepper\u002Flayout styles untouched.\"\n\n**For UI work**, reference the design (a Figma node link, or \"per the design\") rather than\nrestating measurements, and spell out the interaction states: what disables the CTA, the pending\nstate, what happens on success, what renders on error. Name the SCSS partial to create or extend\nand the index file to import it into.\n\nDo **not** include:\n\n- **Rationale.** No \"because\", no \"this keeps X out of Y\", no trade-off discussion, no\n  restatement of a decision's benefits. The design doc and the acceptance criteria hold the\n  *why*; the brief holds the *what*. A reason is the exception, allowed in two places only:\n  where the brief settles a real technical choice an engineer would otherwise stop and\n  question — keep it to a clause — and for a section the issue template asks for that needs\n  nothing (see the next bullet).\n- **Work that isn't needed.** No \"no migration needed\", \"no new settings\", \"no REST route\" for a\n  concern the issue never raised. Silence already says it. The exception is what the template\n  asks for by name — automated tests, and Storybook and VRT changes: when those aren't needed,\n  say so and say why (Step 6).\n- **Routine commands.** No lint, build, test or VRT invocations — the `implement-issue` skill's\n  Step 6 owns verification, and every engineer already knows to build. Mention a command only when it\n  is unusual and specific to this issue.\n- **Links to local design docs.** They don't resolve for someone reading the issue on GitHub;\n  restate the constraint in one clause instead. External links — Figma, a hosted design doc —\n  are fine.\n- **Plain-text line numbers.** No `includes\u002FCore\u002FModules\u002FModules.php:212`, no \"see line 212\", no\n  `file:line` anywhere in the two sections. The number is wrong as soon as anyone edits the file,\n  and a reader on GitHub cannot click it. When a bullet has to point at an exact place in the\n  code, use a GitHub permalink instead — a URL pinned to a commit SHA, with the line or range on\n  the end:\n  `https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002F\u003Ccommit SHA>\u002Fincludes\u002FCore\u002FModules\u002FModules.php#L212-L218`.\n  Get one by opening the file on GitHub and pressing `y`, or with\n  `gh browse --no-browser --commit=\u003Ccommit SHA> \u003Cpath>` and adding the `#L\u003Cstart>-L\u003Cend>` yourself.\n  A URL built on a branch name (`\u002Fblob\u002Fmain\u002F`, `\u002Fblob\u002Fdevelop\u002F`) is **not** a permalink — it points\n  at different lines next week — so never use one. Naming the symbol is still better than linking\n  to it: link only when there is no name to give, such as an unnamed block inside a long function.\n- **Restated acceptance criteria**, a recap of the Feature Description, or background on how the\n  existing system works. This covers the user-facing copy the criteria define — headings,\n  descriptions, CTA labels, error text. Copy gets reworded during brief review, and a verbatim\n  quote in the brief goes stale while the criterion it came from stays current; point at the\n  criterion instead (\"the empty-state heading from the acceptance criteria\"). A string that\n  already exists in the code is not an acceptance criterion — name it exactly, as above.\n- Prose paragraphs.\n\nDo include, when they apply:\n\n- **The right extension on every new frontend file.** New files on the JavaScript side are\n  written in TypeScript: `.ts`, or `.tsx` when the file returns JSX, with co-located tests as\n  `.test.ts` \u002F `.test.tsx`. Never ask for a new `.js` or `.jsx` file. Name the types the new file\n  exports the same way you name any other symbol — the union of allowed values, the interface a\n  table row must match, the function signature with its return type\n  (`classifyContactLink( anchor: HTMLAnchorElement ): ContactLinkType | null`). An existing `.js`\n  file that the issue only edits stays as it is, unless the issue asks for a rename.\n- The concrete shape of any data the change publishes — array keys, JSON payload, inline-script\n  global, selector signature.\n- Changes to existing files that the acceptance criteria imply but do not name.\n\n## Step 6 — Write the Test Coverage\n\nKeep it short — a handful of bullets, not a test plan. Two shapes, both in use:\n\n```\n* Add tests for `\u003Csource file>` covering:\n  * \u003Ccase>\n  * \u003Ccase>\n\n* JS tests in `\u003Cfile>.test.ts` covering the `\u003Cname>` action and the `\u003Cname>` selector.\n\n* Update the tests for each component changed above, so that they check \u003Cbehavior>.\n```\n\n- One bullet per test file or area, with the cases nested under it. Name the test file when it\n  exists or its name follows from the source: co-located next to the source — `*.test.ts` \u002F\n  `*.test.tsx` for every new test file, `*.test.js` only when you are adding cases to a JavaScript\n  test file that already exists — and `*Test.php` under the mirroring path in\n  `tests\u002Fphpunit\u002Fintegration\u002F`. When the same check applies to every file, write one bullet for\n  all of them.\n- Phrase each case as the behavior or outcome, not the mechanics — \"the privacy policy field\n  falls back to the WordPress privacy policy URL when the publication value is missing\", not\n  \"mock the store and assert the input value\".\n- Cover every acceptance criterion at least once, including the negative cases: precondition\n  unmet, feature disabled, invalid input, empty or error response.\n- **Storybook stories belong here, not in the brief.** \"Add a Story for `\u003CComponent>`\", or list\n  the states when there is more than one: \"(default, loading, error)\".\n- **VRT belongs here too, not in a section of its own.** Say what the change does to the VRT\n  scenarios and the reference images — a new scenario for a new story, new reference images for\n  a component whose rendering changes. \"No VRT changes expected\" is a complete answer when nothing\n  the VRT job renders can move. Do not give screenshot paths.\n- **Say so when nothing is needed, and say why.** Never leave Test Coverage empty, and never drop\n  the Storybook or VRT line in silence. When the change needs no tests or no story, write one\n  bullet that says so with the reason — \"No tests required because the change only edits\n  documentation\", \"No Storybook story required because the change adds no UI\". VRT is the\n  exception: a bare \"No VRT changes expected\" needs no reason. An empty section reads as a\n  forgotten section, and a moderator cannot tell the two apart.\n- **Never write `N\u002FA` for tests.** Tests are almost never inapplicable, so `N\u002FA` reads as a\n  skipped question rather than an answer — give the reason instead, as above. The VRT line is\n  where `N\u002FA` is allowed, because there it is often literally true: PHP-only changes, tooling and\n  build config, anything the VRT job never renders.\n- Name any new fixture, fake or mock the coverage needs, and the directory it belongs in.\n- Call out **existing** tests the change will break, so the work is budgeted rather than\n  discovered mid-implementation — \"Fix any failing tests\u002Fsnapshots\" when the blast radius isn't\n  knowable up front.\n- No commands.\n\n## Step 7 — Report what did not go in the brief\n\nEverything you cut for brevity, and everything you learned while verifying, belongs in your\nreply to the user — not in the issue:\n\n- Discrepancies between the design doc and the code, with the `file:line` you checked.\n- Consequences the acceptance criteria did not anticipate: a metric that will shift, a default\n  that changes for every install, an existing test that will start failing.\n- Anything you left as an open decision, and the assumption the brief is written under.\n- Every issue number you were given and what you cited it for, any cross-reference you dropped\n  because no number was available, and any locally-written issue that still has to be filed\n  before the brief's text is final.\n\n---\n\n## Guardrails\n\n- **Write only the two sections.** Leave Feature Description, Acceptance criteria, QA Brief and\n  Changelog entry untouched.\n- **Direction, not steps.** Leave out what the engineer would do anyway; name every area the\n  change touches — see Step 5.\n- **Plain, simple words.** No idioms, no metaphors, no unexplained jargon, no half sentences.\n  Name who does what to what, and list the members of every set you name — see Step 5.\n- **New frontend files are TypeScript.** `.ts` \u002F `.tsx` sources, `.test.ts` \u002F `.test.tsx` tests,\n  and a build-config change whenever the bundle cannot compile them yet.\n- **Names are full words.** Name a new thing only when other code depends on it, in full words\n  — no `Pct`, no `Min`, no initials — see Step 5.\n- **Don't publish.** Do not edit the GitHub issue or post a comment unless the user explicitly\n  asks. Produce the text — or update the local issue file you were given — and let them place it.\n- **Don't expand scope.** If a change looks necessary but no acceptance criterion covers it,\n  raise it in Step 7 instead of adding it silently.\n- **No indirect issue references.** Every cross-reference is a real GitHub number you asked the\n  user for, never a design-doc position, a relative pointer or a placeholder.\n- **No plain-text line numbers.** Every pointer into the code is a GitHub permalink pinned to a\n  commit SHA, never `path\u002Fto\u002FFile.php:212` and never a `\u002Fblob\u002F\u003Cbranch>\u002F` URL — see Step 5.\n- **Verify before naming.** Never reference a symbol, path or hook you have not opened.\n",{"name":5,"description":44,"allowed-tools":45,"argument-hint":46},"Write the Implementation Brief and Test Coverage sections of a Site Kit issue from its acceptance criteria. Use when the user asks to write, draft, adjust, update, or fill in an implementation brief and\u002For test coverage for an issue (e.g. \"write the implementation brief for #12345\", \"fill in the brief and test coverage for this issue\") in the google\u002Fsite-kit-wp repo.\n","Bash, Read, Grep, Glob, Edit, Write","[issue-number-or-path]","develop","b38aae779c9c94c62659315c2c20fa08042e103b",".agents\u002Fskills\u002Fwrite-implementation-brief",391,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp","https:\u002F\u002Fsitekit.withgoogle.com","https:\u002F\u002Fgithub.com\u002Fgoogle","https:\u002F\u002Fopensource.google\u002F","Google ❤️ Open Source","2026-11-14 07:03:26",{"source":8,"skillName":5,"commands":58,"prompt":68,"zipAvailable":69},[59,62,65],{"tool":60,"command":61},"npx","npx skills add google\u002Fsite-kit-wp --skill write-implementation-brief",{"tool":63,"command":64},"bunx","bunx skills add google\u002Fsite-kit-wp --skill write-implementation-brief",{"tool":66,"command":67},"pnpm","pnpm dlx skills add google\u002Fsite-kit-wp --skill write-implementation-brief","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002Fdevelop\u002F.agents\u002Fskills\u002Fwrite-implementation-brief\u002FSKILL.md",true,[],[72,81,86,95,104,112],{"skillNo":73,"name":74,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":39},"01M6991PXQEEZDSJJ0053EK3S8","write-issue-requirements","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002Fdevelop\u002F.claude\u002Fskills\u002Fwrite-issue-requirements\u002FSKILL.md",[],[78,79,21,80],"requirements","issue","write issue requirements",{"skillNo":82,"name":5,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":39},"01M6991NR5S6R94Q2Z6R5SAB1M","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002Fdevelop\u002F.claude\u002Fskills\u002Fwrite-implementation-brief\u002FSKILL.md",[],[22,21,20,19],{"skillNo":87,"name":88,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":39},"01M6991MK67HC2KMG7MM2W8PZ1","write-context-doc","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002Fdevelop\u002F.claude\u002Fskills\u002Fwrite-context-doc\u002FSKILL.md",[],[92,93,21,94],"doc","context","write context doc",{"skillNo":96,"name":97,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":98,"categories":99,"keywords":100,"publishedAt":39},"01M6991K9D80ZZPWRNZZQSHCMP","review-pr","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002Fdevelop\u002F.claude\u002Fskills\u002Freview-pr\u002FSKILL.md",[],[101,102,103],"pr","review pr","review",{"skillNo":105,"name":106,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":39},"01M6991J07Q5HB2ZCRD2JRWR7Z","implement-issue","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002Fdevelop\u002F.claude\u002Fskills\u002Fimplement-issue\u002FSKILL.md",[],[79,110,111],"implement","implement issue",{"skillNo":113,"name":114,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":115,"categories":116,"keywords":117,"publishedAt":39},"01M6991GS58C967HP03TW1PMEJ","check-pr-failures","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fsite-kit-wp\u002Fblob\u002Fdevelop\u002F.claude\u002Fskills\u002Fcheck-pr-failures\u002FSKILL.md",[],[118,119,101],"check pr failures","check"]