Update to MakeBox for Etsy

{
  "description": "Research product-relevant Etsy keywords, competitors and niches from seller reports, CSVs, public samples or MakeBox estimates; explain sources, missing metrics and selection reasons without inventing demand or ranking guarantees.",
  "included_files": [],
  "name": "research-etsy-market",
  "skill_md_contents": "---\nname: research-etsy-market\ndescription: Research product-relevant Etsy keywords, competitors and niches from seller reports, CSVs, public samples or MakeBox estimates; explain sources, missing metrics and selection reasons without inventing demand or ranking guarantees.\n---\n\n# Research keywords and niches\n\nRead [research guide](../../references/keyword-research.md). First establish item, real traits, buyer intent, language/market and scope; use [brief](../../references/product-brief.md) only for missing relevant facts.\n\n## Workflow\n\n1. Gather seller search/conversion reports or supplied datasets with date/definitions. MCP absent: analyze these locally or use authorized public browsing. With no numeric evidence, provide clearly labelled semantic hypotheses.\n2. Build intent groups and reject misleading materials/product types, unrelated holidays and unauthorized brands before ranking candidates.\n3. Connected: `search_keywords` finds candidates (up to100); `check_keywords` validates batches of at most40. `competitor_tags`/`search_top_listings` give sample language/offers. Inspect each current schema; direct public alternatives are in the [index](../../references/etsy-operation-index.md).\n4. With thousands of CSV terms, filter/deduplicate and shortlist rather than making thousands of calls. Missing statistics remain unknown. Compare relevance, intent and coverage before volume/competition.\n5. Compare like-for-like offers, not a single cup versus a set or a template versus print. Shop lifetime sales are not listing sales; returned search order is not universal rank.\n6. Deliver the [evidence table](../../templates/keyword-evidence.md), recommended primary phrase, title/tag allocation, rejected misleading phrases and a feasible test plan. Tags fit ≤20 characters; natural longer phrases may remain in prose.\n\nRetain supplied clicks/trends/seasonality only if actually present and defined. Provider opportunity is not a conversion probability. Never claim exact search demand from tag frequency, guarantee top placement, copy protected content or request private competitor orders. Research alone does not authorize any shop change or paid internal generation.\n"
}