- Home
- Skills
- E-commerce
- Product Feed Doctor for Google Merchant Center (and Meta Cat
Product Feed Doctor for Google Merchant Center (and Meta Cat
**What it checks, per item:** - **Required attributes**: id, title, description, link, image_link, availability, price, plus the brand / gtin / mpn rules and condition.
$2
Works with the AI Agents you already use
Product Feed Doctor for Google Merchant Center (and Meta Cat
Example session with this skill installed
Check our public demo store feed https://storefront.e-com.plus/products-feed.xml?store_id=1011&domain=storefront-demo.e-com.plus — why would products be disapproved in Merchant Center? Also test the images.
- Read your context and instructions
- Compiled the product feed doctor
Real run on 23 Oct 2026: scripts/feed_doctor.py --check-images on the live public demo-store feed of the E-Com Plus platform (Atom, gzip, 133 entries). Shortened. The feed is live, so your numbers may differ.
Product feed check: storefront-demo.e-com.plus
Checked 23 Oct 2026 · 133 items · Atom · Google Merchant Center
8 of 133 items (6%) will likely be disapproved. 79 more have warnings · 46 look clean.
Image check: 33 of 33 unique image URLs requested, 0 broken.
Fix first
1.
GTIN check digit wrong, 5 items, e.g. four notebook variants (7908387300469, 7908387300480…) and another notebook (0000980980980). Copy the GTIN from the barcode or remove it; a wrong GTIN is worse than none.
2.
Missing price (AVI6034 "kit teste"),
missing image_link (YSX9427),
missing brand (12349). The first two look like test products, so exclude them from the feed.
3. No GTIN and no MPN on 52 items (warning). Only 12 of 133 items (9%) have a gtin, and only 64 (48%) have a brand.
| Severity | Issue | Items |
|---|---|---|
| ERROR | GTIN check digit wrong | 5 |
| ERROR | Missing image_link / brand / price | 1 each |
| WARNING | No GTIN and no MPN | 52 |
| WARNING | Variants in a group don't differ (e.g. 23 phone cases in one group with no color/size/pattern difference) | 26 |
| WARNING | Same main image on different products | 11 |
| WARNING | Same title on several items, no item_group_id | 8 |
| WARNING | Description over 5000 characters (7603) | 1 |
| INFO | availability written as in stock (spec lists in_stock) | 133 |
| INFO | No shipping attribute: fine if set in Merchant Center account settings | feed |
Coverage, why some products underperform: google_product_category 2% · gtin 9% · brand 48% · color 53% · size 35% · gender and age_group 0%.
Inside product_type "Notebooks gamers" (30 items): gtin 4/30, brand 12/30, size 15/30, item_group_id 17/30, color 19/30.
Not checked: policy reviews, landing pages, price match with the site, account shipping/tax settings, image pixel size. Only Merchant Center's Diagnostics is authoritative.
Second real test, same day: a 2019 snapshot of a European wine & deli shop's Google feed (public in a validator's test folder):
19 of 19 items will likely be disapproved, one root cause: every price uses a decimal comma (169,00 EUR), so one setting in the feed app fixes all 19. It also found sale_price sent as just EUR, google_product_category as free text instead of a Google taxonomy value, and one GTIN with a wrong check digit.
Connects securely to your tools. The creator never sees your data.
About this skill
Give your agent a feed URL or file and ask: "Why are half my products disapproved?" It works with RSS 2.0 / Atom XML using the g: namespace, and with tab-delimited or CSV files, whatever generated them: Shopify apps, WooCommerce plugins, Magento, PrestaShop, a spreadsheet export or your own code.
What it checks, per item:
- Required attributes: id, title, description, link, image_link, availability, price, plus the brand / gtin / mpn rules and condition. Each rule was checked against Google's own Merchant Center help pages on 23 Oct 2026, and the skill cites them.
- Formats: price + ISO 4217 currency (
15.00 USD, not$15or15,00 EUR), availability values, title (150) and description (5000) length, duplicate and case-only duplicate ids, invalid or relative URLs, http images, placeholder and reused images. - Identifiers: GTIN length, GS1 check digit and restricted / coupon prefixes, placeholder brands ("N/A", "Generic"),
identifier_existsconflicts. - Consistency: sale_price not lower than price, preorder without availability_date, mixed currencies, links on several domains, attributes written without the
g:prefix (Google ignores them). - Apparel and variants: color / size / gender / age_group for apparel, same-title items without item_group_id, and variant groups whose items don't differ.
- Optional image check: one HEAD request per unique image URL (up to 100) finds broken photos and URLs that don't return an image.
Coverage table: the "half my products don't show" answer. It shows the share of items with google_product_category, product_type, brand, gtin, mpn, color, size, gender, age_group, item_group_id and more, overall and inside each category. "15 of 30 notebooks have no size" is visible at a glance.
You get a short summary with the root cause first (one feed-app setting often breaks every item), a table of issues with the exact fix and a before → after for each, and a per-item table with errors first. The helper also writes the full result as JSON.
Honest by design. ERROR = required value missing or invalid, so the item will likely be disapproved; WARNING = truncated, limited, or inconsistent; INFO = allowed or not verifiable from the file. Rules the skill could not verify are listed as unverified, not stated as facts. Every report ends with the limit: only Merchant Center's own diagnostics are authoritative. Policy reviews, the landing-page crawl, price mismatches with your site and account-level shipping/tax settings are not visible in a feed file.
Read-only: one GET for the feed, optional HEAD requests for images. No logins, no uploads, no changes to your store.
- Feed file only: no policy checks, landing-page crawl, price/availability match with your website, or account-level shipping, tax and return settings.
- Image check: reachability and content type only. It does not measure pixel size (Google's 500 x 500 px minimum applies from 31 Jan 2027) or check robots.txt for Googlebot-image.
- No full Google taxonomy: apparel is recognised from category IDs 166, 1604 and 187 or the "Apparel & Accessories" path.
- Tax isn't checked. Google's tax page couldn't be verified on 23 Oct 2026.
- Rules were verified on 23 Oct 2026. Google updates the specification several times a year, so re-check the linked pages if Merchant Center disagrees.
- It doesn't fix the feed or the store. It gives exact changes for you or your developer.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 25 days ago
Requires
Permissions
Allowed Hosts
Network: one HTTPS GET for the feed URL you give. With the image check, one HEAD request per unique image URL (a 1-byte GET if HEAD is refused), 6 at a time, at most 100 by default. There are no other hosts, no telemetry and no API keys. The helper sends an honest user agent ("product-feed-doctor … feed owner self-check; read-only"). It writes files only where you point --json / --md. It never logs in to Merchant Center, Meta or your store, never uploads or edits feeds, and refuses XML that declares entities. The package is text only: Markdown and one Python script, no secrets.