Skip to content

AgriciDaniel/claude-blog / skills/blog-google

blog-google

Verified

Credentialed Google API calls for a blog: PageSpeed Insights, CrUX history, Search Console performance and URL Inspection, GA4 organic traffic, Cloud NLP entities, YouTube video search, Google Ads Keyword Planner, and the Indexing API (job and livestream pages only). Uses the user's Google credentials, quotas, and sometimes billing, and asks for consent before the first call. Runs only on an explicit request: /blog google, "run PageSpeed on this blog URL", "pull my Search Console data", "inspect this URL's index status", "set up Google API credentials". Shares config with claude-seo at ~/.config/claude-seo/google-api.json.

references/search-currentness.md

View on GitHub
# Google Search Currentness Playbook

Verified against Google-owned sources on 2026-10-11. The ledger was reviewed
again on 2026-10-29 for Search Central documentation updates through
2026-10-26; sections below that this review did not touch keep the earlier
date. Treat product announcements as product context, not evidence of a
ranking factor. Use `data/google-updates.json` for the machine-readable source
ledger.

Resolve the ledger in this order:

1. Resolve the main blog skill from the trusted installed sibling skill root
   supplied by the caller. Ordinary reference files do not receive host path
   substitutions themselves.
2. If that skill's installed ancestor contains `.claude-plugin/plugin.json`,
   use `../../data/google-updates.json` relative to the main skill. This covers
   both repository and cached plugin layouts.
3. Otherwise use `data/google-updates.json` beside the main blog skill, normally
   `~/.claude/skills/blog/data/google-updates.json` for a standalone install.

Do not substitute a same-named file from the current working directory. If
the selected trusted location is missing, stop ledger-backed currentness work
and report it as unavailable. Primary-source research can continue with an
explicit evidence gap; it cannot establish that the missing ledger is current.

The ledger separates four evidence states:

- `confirmed`: the cited Google source directly supports the recorded fact.
- `confirmed-event-pending-impact`: the event exists, but its site impact and
  targets are not established.
- `confirmed-data-anomaly`: Google identifies a reporting defect, not a ranking
  change.
- `confirmed-with-source-conflict`: current Google-owned pages disagree. Keep
  both citations and verify the capability in the affected account.

Run `python3 "${CLAUDE_BLOG_SCRIPTS_DIR:-${CLAUDE_SKILL_DIR}/../../scripts}/check_google_currentness.py" --root "${CLAUDE_SKILL_DIR}/../.." --json`
(resolve the directory first and reject it unless absolute; never run a
relative scripts path) for the read-only official-feed and review-age check. A `refresh_required` result means
the ledger needs human review. The checker never writes summaries or promotes
events automatically.

## Core and Spam Update Analysis

The latest confirmed ranking event is the September 2026 spam update. Google
says it began on September 24, applies globally and to all languages, and may
take up to two weeks to complete, and its status dashboard marked the rollout
complete as of October 8. Google did not publish a target profile. Collect one
full post-rollout week (October 15 or later) before comparison. Until then,
record impact as `PENDING_OBSERVATION`.

1. Confirm the named update's start and end on the Search Status Dashboard.
2. Wait at least one full week after completion.
3. Compare a post-update week with a week before rollout.
4. Analyze Web Search, Images, Video mode, and News tab separately.
5. Distinguish small movement from a large, sustained drop.
6. Avoid quick fixes and mass deletion. Improve reader value and navigation in
   durable ways; deletion is a last resort.
7. Do not infer what an update rewarded from dates alone. Smaller core changes
   may be unannounced.

## Search Console Data Anomalies

Check Google's data-anomalies page before attributing a reporting change to a
ranking update. Google records these August 2026 logging defects:

- August 13: lower reported Discover clicks and impressions, including lower
  generative-AI Discover impressions for properties with that report.
- August 13 through August 17: lower reported impressions in the generative-AI
  Search report.

These defects affect logging only. They begin before the August 18 spam update,
so their dates cannot be used as evidence that the spam rollout caused an
August 13 through August 17 decline. Google's August 21 update says the missing
generative-AI report data was restored. Do not describe it as an ongoing loss.

## Canonical Reevaluation

After a material canonicalization fix, Google may keep the URL in the duplicate
cluster for up to two weeks. Report the state as `PENDING_REEVALUATION` during
that window when the implementation is now correct. Search Console's Request
Indexing feature is quota-limited; reserve it for important URLs.

Request Indexing in URL Inspection is separate from the Indexing API. The
Indexing API remains restricted to eligible JobPosting and
BroadcastEvent/VideoObject pages.

## October 2026 Documentation Updates

Documentation changes, not ranking events. Cite the ledger entries for detail.

- URL structure (October 23): hyphens, not underscores, between words; readable
  words over ID numbers; URLs are case sensitive; no slug length limit is given.
- Control what you share (October 21): one page now lists noindex, robots.txt
  for images and video, robots meta, `data-nosnippet`, `X-Robots-Tag` and a
  site-wide Search generative AI control in Search Console. The page does not
  describe how that control works, so do not claim a mechanism.
- User-generated spam (October 26): consider `noindex` for posts from new
  users, `nofollow` or `ugc` on links in untrusted content, and watch the Page
  Indexing report for spikes. Relevant to comment sections.
- Indexing API quota (October 16): 200 publish requests per day, 180 metadata
  requests per minute, 380 requests per minute overall. Eligibility is
  unchanged, so blog posts are still not eligible.
- UGC Fresh Data Program (October 8): application-only pipeline for platforms
  that primarily host user-generated content. Not relevant to ordinary blogs.

## Generative AI Performance Reports

Google announced the dedicated Search Console generative-AI views on June 3.
The announcement's August 31 update says they have rolled out to all websites
worldwide. Check the account for its actual available data:

- Separate Search and Discover reports.
- Search includes AI Overviews and AI Mode.
- Documented dimensions are impressions, pages, countries, devices for Search,
  and dates.
- Do not promise clicks or queries in these dedicated reports.
- No supported Search Console API endpoint is documented for this dedicated
  view. The blog-google command must report `SKIPPED` or unavailable and direct
  the user to the Search Console UI rather than synthesize data.

Standard Search Analytics totals continue to include AI-feature activity under
Google's documented aggregation rules; do not claim those totals isolate AI
Overviews or AI Mode.

## Platform Properties

Google's July 29 Search Central announcement says platform properties are
globally available to everyone for Instagram, TikTok, X, and YouTube. The
current Search Console Help page still says the feature is rolling out
gradually. Treat availability as `SOURCE_CONFLICT`: check the actual account,
cite both Google-owned pages, and do not promise that the current Search Console
API or `/blog google gsc` supports these reports.

## Discover

Run this checklist only when Discover is a declared target or the property has
Discover data:

- Useful, original, in-depth, and timely material.
- Country and topic relevance where applicable.
- Non-sensational titles and non-clickbait presentation.
- Topic-level expertise; older useful content remains eligible.
- No special structured data requirement.

Preferred images are at least 1200px wide, contain more than 300,000 total
pixels, use a useful 16:9 crop where possible, and are enabled by
`max-image-preview:large` or AMP. Use a relevant, representative image through
schema.org markup or `og:image`.

## Preferred Sources

Preferred Sources is optional audience development, not a ranking signal. It
works at domain or subdomain level, not subdirectory level. For a user who
selects the publication, its content is more likely to appear in Top Stories
and can receive a preferred badge in AI Mode or AI Overviews. Offer Google's
publisher assets only when this fits the site's audience strategy.

Google's August 20 documentation supports standard and custom interactive
buttons plus a non-JavaScript deeplink. The choice must remain reader-triggered.
Do not describe implementation as a general ranking improvement.

## Review Snippet Integrity

Google's July 24 review guidance prohibits fake reviews and incentivized reviews
without clear and prominent disclosure. Before recommending Review or
AggregateRating markup, verify all of the following:

- The review is based on a genuine experience and was not fabricated.
- Any benefit, payment, discount, voucher, or free product is disclosed clearly
  and prominently.
- Review text and ratings in structured data are visible on the page.
- Aggregate ratings are not copied from other websites.

Parsing success cannot establish authenticity. If the evidence is unavailable,
report the review as unverified and do not generate a review or rating.

## Google Ads API Currentness

Google Ads API v25.2 was released on 2026-09-23. Check Google's release notes
and support table together before changing a versioned path or client library.
The repository's existing Python client constraint may lag the current client
release. Dependency updates require an offline compatibility test for Keyword
Plan services and requests before a live, credentialed call. Never use a
developer token, enable billing, or run a live Ads request merely to prove
package compatibility.

## Crawl and Interaction Checks

- Googlebot processes the first 2MB of a supported file and the first 64MB of a
  PDF, measured uncompressed. Place critical title, metadata, canonical,
  essential schema, and primary content before the HTML cutoff.
- Warn on inline base64, CSS, JavaScript, or navigation bloat that can push
  critical content beyond the first 2MB. This is not a ranking factor.
- Back-button hijacking requires observed deceptive behavior. Do not flag
  normal History API use by syntax alone.
- A section intended for a "Read more" deep link should be immediately visible
  and retain its hash on load. Do not force a scroll reset. This is not a ban
  on every accordion elsewhere.

## AMP

AMP is supported, not required, and has no special ranking benefit. Since
2026-07-01 Google Search sends users directly to publisher-hosted AMP pages.
Keep AMP only when it provides operational value; remove it with correct
canonicals and redirects.

## Generative AI Product Context

Google I/O 2026 announced Gemini 3.5 Flash as AI Mode's global default,
follow-ups from AI Overviews into AI Mode, multimodal inputs, and information
agents. The May Explore-the-web update highlighted original analyses, public
discussions, inline links, and link previews.

These announcements do not create new content-scoring factors. Do not recommend
agent-specific schema, fixed-size content chunks, or fan-out page factories.
Use foundational SEO, accurate non-commodity material, authentic discussion,
clear page identity, and useful media.

## Primary Sources

- https://developers.google.com/search/updates
- https://developers.google.com/search/updates/search_docs_updates.rss
- https://status.search.google.com/incidents.json
- https://developers.google.com/search/docs/fundamentals/using-gen-ai-content
- https://developers.google.com/search/docs/appearance/structured-data/video
- https://developers.google.com/search/docs/appearance/regional-search-experiences
- https://status.search.google.com/incidents/LEubPCm2octf2uMqCFKE
- https://status.search.google.com/products/rGHU1u87FJnkP6W2GwMi/history
- https://support.google.com/webmasters/answer/6211453?hl=en
- https://developers.google.com/search/docs/appearance/core-updates
- https://developers.google.com/search/docs/crawling-indexing/canonicalization-troubleshooting
- https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
- https://developers.google.com/search/blog/2026/06/gen-ai-performance-reports
- https://developers.google.com/search/blog/2026/07/platform-properties-social-video-guide
- https://support.google.com/webmasters/answer/17148418?hl=en-GB
- https://developers.google.com/search/docs/appearance/google-discover
- https://developers.google.com/search/docs/appearance/preferred-sources
- https://developers.google.com/search/docs/appearance/structured-data/review-snippet
- https://developers.google.com/google-ads/api/docs/release-notes
- https://developers.google.com/google-ads/api/docs/sunset-dates
- https://developers.google.com/search/docs/crawling-indexing/googlebot
- https://developers.google.com/search/blog/2026/04/back-button-hijacking
- https://developers.google.com/search/docs/appearance/snippet
- https://blog.google/products-and-platforms/products/search/search-io-2026/
- https://blog.google/products-and-platforms/products/search/explore-web-generative-ai-search/
receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementationStars: 322kUpdated Claude CodeCodexVerifiedInstall
requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirementsStars: 322kUpdated Claude CodeCodexVerifiedInstall
using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallbackStars: 322kUpdated Claude CodeCodexVerifiedInstall
verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions alwaysStars: 322kUpdated Claude CodeCodexVerifiedInstall