Every Product Hunt launch, maker and trend, queryable.

119,743 individual launches, one record per product, as of the 2026年9月6日 snapshot. Every record carries its topics, upvotes, ranks and launch history from ingestion; description, website, pricing type, lifecycle state and parsed launch year fill in as a separate hydration pass works through the archive, so a newly-archived launch can be missing those fields temporarily.

Entirely from Product Hunt's own public launch archive, pages and profiles anyone can browse without an account — no login, no API key, no paid Product Hunt developer program involved. The value is that it's cleaned, deduplicated and queryable as one index instead of one page at a time.

producthunt-makers is a separate index of 95,111 public maker profiles — the people who made a launch — each carrying a products-made / total-upvotes / topics footprint for leaderboard-style queries. It cross-references the products dataset two ways: filter products by maker={username} to see everything one person made, or read the makers[] field on a product record for who made it.

producthunt-trends rolls up launch counts, total/average upvotes and average rating by topic, optionally within a calendar month or year — 10,629 cells at the default monthly granularity. It never returns an individual product or maker record and has no /items endpoint, because a trend cell isn't an entity — it's a rollup bucket. Small cells (currently fewer than 3 launches) are suppressed entirely rather than shown with a misleadingly precise average.

Those fields come from a hydration pass that works through the 119,743-product archive over time, separately from the initial launch/topic/upvote record. Different fields reach different shares of the archive at different rates — website links reach the widest slice, a parsed review rating the sparsest. Absence means "not hydrated yet", not a data-quality gap; re-query later or filter on has_website / is_online / min_rating to work with the populated slice directly.

Yes. GET /datasets/producthunt-products/items/{slug} returns one launch's full record including its complete launch history; GET /datasets/producthunt-makers/items/{username} returns one maker's full profile and the products they made. producthunt-trends has no item lookup — query its search or facets endpoint for a topic/period cell instead.

Every trend cell (a topic, optionally within a period) is dropped entirely unless it clears a minimum launch-count floor — currently 3, and min_launches can only raise that floor further, never lower it. This keeps every reported average_votes or avg_rating backed by enough launches to be meaningful, rather than one product swinging a two-launch cell's average wildly.

Yes, across all three datasets. Products: search, items/{slug}, facets. Makers: search, items/{username}, facets. Trends: search and facets (no items). All eight are exposed as MCP tools (datasets_producthunt_products_search, datasets_producthunt_products_item, datasets_producthunt_products_facets, datasets_producthunt_makers_search, datasets_producthunt_makers_item, datasets_producthunt_makers_facets, datasets_producthunt_trends_search, datasets_producthunt_trends_facets) and billed pay-on-success.