agentleFS
SpytrendSpytrend/spytrend-mcp/skills/spytrend-shop-discovery/SKILL.md

spytrend-shop-discovery

Skill30 starsChanged 3 months agoMITView on GitHub

Discover store domains by traffic, qualified growth and advertising activity with SpyTrend MCP data, then inspect a bounded set of leading stores' ads and advertisers, from a database of 1B+ tracked Meta ads. Use when the user wants to find growing e-commerce or dropshipping stores, see which shops advertise most in a niche or country, run product research through stores that are taking off, or scout new stores that just started running heavy ads. Triggers: "shop spy", "fast-growing Shopify stores", "find winning products", "product research", "stores scaling their ads"; RU «найди быстрорастущие магазины», «товарка: какие магазины взлетают», «спай магазинов и товаров»; ES «productos ganadores»; PT «produtos vencedores». Not for one known domain (use spytrend- website-review) and not for advertiser rankings without stores (use spytrend- advertiser-overview).

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 3 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

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

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

    Read references/spytrend-contract.md before making any data call.

    line 20
  2. Distinctive:

    Default: five admitted rows per cut, up to three store drilldowns, chat output, media off.

    line 27
  3. Very specific:

    Call get_usage, state the three page limits, drilldown cap, media choice and worst-case spend.

    line 32
  4. Very specific:

    Call search_shops three times with the same filters and separate server sorts: traffic, qualified growth, and advertising activity.

    line 38
  5. Very specific:

    When the cut comes back empty, say that no store currently qualifies and give the snapshot month from traffic_as_of as the reason.

    line 43
  6. Distinctive:

    If filtering leaves fewer rows than requested, page forward with the returned offset only until the target is filled or three pages have …

    line 49
  7. Distinctive:

    Use the server's selected-window creative reuse ordering when available.

    line 59
  8. Very specific:

    Call advertiser cards only for unique returned advertiser ids and search_webmasters by exact domain with a small limit for …

    line 61
  9. Very specific:

    Use references/report-template.md or the packaged renderer with references/report-template.html.

    line 74

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

SpyTrend shop discovery

Read references/spytrend-contract.md before making any data call.

Inputs

  • Require at least one of vertical or ISO-2 country.
  • Platform is optional. Omit the parameter for "any platform"; never send a made-up wildcard value.
  • Default: five admitted rows per cut, up to three store drilldowns, chat output, media off.

Preflight

Call get_usage, state the three page limits, drilldown cap, media choice and worst-case spend. Decide whether domain scope is exact host or root plus subdomains and label it.

Workflow

  1. Call search_shops three times with the same filters and separate server sorts: traffic, qualified growth, and advertising activity. Never re-sort one bounded page to manufacture another cut. The growth cut returns only stores whose growth estimate the service has qualified, and that qualification depends on how fresh the traffic snapshot is. When the cut comes back empty, say that no store currently qualifies and give the snapshot month from traffic_as_of as the reason. Render the other two cuts and continue. Never substitute a re-sorted traffic page for it, and never report an empty qualified cut as "no growing stores".
  2. Apply the public quality fields returned by the service. Reject infrastructure and redirect surfaces. Preserve server order among admitted rows.
  3. If filtering leaves fewer rows than requested, page forward with the returned offset only until the target is filled or three pages have been inspected. Report fetched, admitted and rendered counts. Never promise implicit top-up.
  4. Traffic is a monthly estimate for traffic_as_of. Growth is shown only when qualified. ads_monthly counts ads first discovered by SpyTrend; it is not a platform-launch series, and the open month is incomplete.
  5. Select at most one unique admitted store from each cut for drilldown. For each, use search_ads with landing_domain_exact=true, the exact domain and concrete dates for a bounded count and representative ads. Use the response's total status and date axis.
  6. Use the server's selected-window creative reuse ordering when available. Keep window reuse and lifetime reuse separate.
  7. Call advertiser cards only for unique returned advertiser ids and search_webmasters by exact domain with a small limit for advertising-network attribution. A card without a card_status field is complete; card_status=building means its counters are unavailable, not zero.
  8. With confirmation, fetch media only for the representative ads in the final drilldowns.

Report

Show each cut independently, monthly estimate period, first-discovery label, quality admission, fetched/admitted coverage, store drilldowns, unavailable sections and spend. A domain surface is not automatically a brand or company.

Use references/report-template.md or the packaged renderer with references/report-template.html; otherwise provide the complete result in chat.

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.