agentleFS
jbobbinkjbobbink/seo-gsc-wizard/skills/merchant-center-health/SKILL.md

merchant-center-health

Skill1 starsChanged 8 days agoMITView on GitHub

Check whether the Merchant Center account linked to a Search Console property is healthy and whether its products are eligible for free listings, using the GSC Wizard MCP Merchant Center tools: readable accounts and links, account issues by severity, feed data sources with their latest uploads, and the products invisible in free listings with the reason, ending in a verdict and recommendations. Use when the user asks "is my Merchant Center account ok", "why are my products disapproved", "are my products showing in free listings", "is my feed uploading", "Merchant Center health check", or mentions a suspension, a policy issue or a feed that stopped.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 0 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Resolve the property: If the user's request or the conversation names a property, use it.

    line 12
  2. Very specific:

    Find the linked account: Call list_gmc_accounts with the property's siteUrl (the argument is optional, but passing it adds linkedAccountId, …

    line 14
  3. Very specific:

    Account health: Call get_gmc_account_health with siteUrl, optionally languageCode (BCP-47, for example "nl", default en-US) for the issue …

    line 20
  4. Very specific:

    Feed data sources: Call get_gmc_data_sources with siteUrl (includeUploads defaults to true and probes the latest upload of up to 25 file …

    line 26

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Merchant Center Health

Check the health of the Merchant Center account linked to one Search Console property and find the products that are not serving in free listings, ending with a verdict and a fix list. The user's request may contain a property, a Merchant Center account name or id, a language code for the issue text, or a focus hint (for example "only the feed" or "disapproved products"); honor the hint by going deeper there, but still run every step.

Workflow

  1. Resolve the property. If the user's request or the conversation names a property, use it. Site URLs use GSC formats: sc-domain:example.com or https://www.example.com/. Otherwise call list_sites and pick the obvious match. If several plausible properties exist, ask the user which one before continuing.
  2. Find the linked account. Call list_gmc_accounts with the property's siteUrl (the argument is optional, but passing it adds linkedAccountId, the account this property is linked to). The accounts[] rows carry accountId, accountName, accountEmail, isAdvanced and linkedSiteUrls. Interpret the response before anything else:
    • connected: false with reason: "no_scope" means the Merchant Center consent has not been granted in the GSC Wizard app; reason: "credentials_revoked" means it was granted but Google now rejects the saved connection, so the Google account has to be reconnected on the Account page. Relay the hint and stop the Merchant Center steps, since nothing below can answer.
    • Accounts listed but no linkedAccountId means the property is not linked yet. Linking happens in the GSC Wizard app, not through a tool: show the account table and tell the user which account to link. For a grantee without the Merchant Center opt-in the id is withheld with a note; say so.
    • isAdvanced: true marks a multi-client parent that cannot be reported on; one of its sub-accounts has to be linked instead.
    • incomplete: true means the list is a floor: read gaps, unprobedAccountIds and errors, say what could not be read, and suggest a retry.
  3. Account health. Call get_gmc_account_health with siteUrl, optionally languageCode (BCP-47, for example "nl", default en-US) for the issue text and limit (default 200, max 500, most severe first). Read overallStatus first: it is the worst of the checks and the issues. Then read:
    • issues[] (title, detail, severity, affectsFreeListings, impactedDestinations with the reportingContext and the regions each one hits, documentationUri) and the issueCounts totals.
    • checks[] (key, status, message), where status is one of ok, suggestion, warning, critical or unknown. A setting this Google login may not read answers unknown and sections says why: report it as unverified, never as passed.
    • programs[] for the free-listings program state, its activeRegionCodes and any unmetRequirements; homepage.claimed for whether the website is claimed; countriesWithoutShippingService against targetCountries for the markets without an account-level shipping service. Business info is reported as presence flags only (businessInfo), and shipping and returns set per product in the feed are not read, so those two checks can flag what the feed already covers.
  4. Feed data sources. Call get_gmc_data_sources with siteUrl (includeUploads defaults to true and probes the latest upload of up to 25 file sources, primary first; set it to false to list the sources only). Read:
    • findings[] (displayName, severity, message) first: failed uploads, files processed with errors, fetches that stopped running and primary sources with free listings switched off, with failedUploadCount and staleFetchCount as the totals.
    • rows[]: type, input, feedLabel, contentLanguage, countries, destinations with their state, uploadStatus (not_file_input for API, UI and autofeed sources, which have no uploads), latestUpload with processingState, uploadTime, itemsTotal, itemsCreated, itemsUpdated and issues, latestUploadAgeDays, and staleFetch, which compares the newest processed file with the fetch schedule (2, 9 or 35 days for daily, weekly or monthly). Fetch URLs are reduced to scheme and host on purpose. A stale or failed primary source explains a catalog-wide drop before any product-level issue does.
  5. Invisible products. Call get_product_listing_status with siteUrl, onlyProblems: true (the default) and limit (default 25, max 200). Rows come sorted by click potential, clickPotentialRank 1 being the product Google expects the most clicks from, so a disapproved product with a low rank number is the most expensive problem. Read problemCount and merchantActionCount for the whole-catalog counts (they are not page counts), statusCounts and issueCounts[] (code, resolution, count) for the mix, and per row offerId, title, aggregatedStatus, freeListingsStatus, topIssueCode, topIssueSeverity and topIssueResolution, where MERCHANT_ACTION needs a feed or site fix and PENDING_PROCESSING clears on its own. When hasMore is true, pass nextOffset back as offset only if the user wants the whole list; the counts already cover it. Read the freshness and provenance flags every time:
    • dataSource: "clickhouse" is the daily catalog snapshot, syncedAt says when and stale: true means it is more than two days old. dataSource: "merchant-api" is a live scan of up to 5,000 products, and truncated with truncationReason says when the catalog had more or the read ran out of time.
    • itemIssuesUnavailable: true means Google refused the issue data, so a missing issue code is unknown, not clean.
    • syncPaused (reason: "merchant_access_lost", with a one-line staleNote) means the daily sync is paused because Merchant Center no longer lets the property's Google account read the merchant; the stored data is still returned but may be stale, and the message says how to fix it.
    • syncNotice (reason: "gmc_on_other_account") means Merchant Center is connected on another Google login than the property's Search Console one, the only account the sync reads through, so the stored snapshot cannot refresh; with nothing stored and no live read possible the tool answers notConfigured with that reason instead.
  6. Interpretation rules. Judge severity by the clicks at stake: a critical account issue that affects free listings or a failed primary feed outranks any number of suggestion rows, and a disapproved product near the top of the click-potential ranking outranks a dozen low-potential ones. Present every multi-row result as a compact markdown table (top rows only, typically 5 to 10); keep prose for interpretation, not for restating numbers. Do not fabricate rows the tools did not return; if a tool returns nothing, say so and quote its emptyNote when there is one.

Output

Present the findings in this order:

  1. Account verdict. One short paragraph plus overallStatus from get_gmc_account_health, naming the linked account (linkedAccountId and its accountName from list_gmc_accounts) and the data freshness of the catalog snapshot (syncedAt, stale, syncPaused, syncNotice).
  2. Issues by severity. From get_gmc_account_health, most severe first, followed by the eligibility checks that are not ok, with every unknown one marked as unverified:
SeverityIssueAffects free listingsDestinations / regionsFix (documentationUri)
  1. Feed status. From get_gmc_data_sources: one row per source with type, input, latest upload or fetch time, processing state, items and age, then the findings in prose.
  2. Invisible products. From get_product_listing_status: the whole-catalog problemCount and merchantActionCount, the issueCounts mix, and the top rows by click potential:
RankOfferTitleStatusTop issueResolution
  1. Recommendations. End with a short prioritized list of 3 to 7 concrete recommendations, ordered by clicks at stake: resolve the account issue or claim the website, fix or re-upload the failing primary feed, correct the top MERCHANT_ACTION issue codes on the highest-potential products, and wait out the PENDING_PROCESSING ones. When the question drifts to feed attribute quality or product titles, point the user to the Feed Audit and Title Hygiene reports in the app; for how the visible products perform, run the shopping-performance skill.

Error handling

If any tool call fails with an authentication or subscription error, tell the user to run /mcp to (re)authenticate, and that a GSC Wizard account with an active plan or trial plus a connected Google Search Console account is required. If the property is not found, re-run list_sites and offer the closest matches instead of guessing a URL format. If a single tool fails for another reason, report the gap in the relevant section and complete the rest of the workflow with the remaining data.

Merchant Center is an optional connection, so every tool in this workflow answers notConfigured: true with a reason rather than erroring, and the reason decides what to tell the user:

  • no_scope: Merchant Center is not connected in the GSC Wizard app yet.
  • credentials_revoked: the consent exists but Google rejects the saved connection; reconnect the Google account on the Account page.
  • no_account: no Merchant Center account is linked to this site; show the list_gmc_accounts table and say which one to link in the app.
  • gmc_on_other_account: the scope sits on another Google login than the property's Search Console one, so the warehouse cannot fill and nothing is stored yet.

A merchant that refuses every connected Google account fails with forbidden and fixed wording that says access to it was lost and how to restore it; relay that wording as the verdict and stop the Merchant Center steps.

Grounding

Every number you report must name the tool it came from and the date range it covers, for example "clicks 12,340 (get_site_summary, 2026-09-21 to 2026-09-27)". For a table, state the tool and date range once on the line above it. If you derive a figure yourself (a delta, a sum, a share), say so and name the tool outputs it was computed from. Never report a number no tool returned.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.