Item level, crawled nightly
crawl 4 days agoSearch the index
75,944 items985 registries34,217 verified free11,020 verified paid30,707 unverified18,895 previewable live11,371 removed in 30 daysrebuilt 2026-09-13 · captured 2026-09-13

Best shadcn registries

ranked on four published facts, 624 eligible

624 registries publish 8 or more components and are still serving a manifest. These are the 25 that score highest on four published facts.

Start at shadcn's own registry directory, because it costs nothing and the registries in it are "built into the CLI with no additional configuration required", so `npx shadcn add @<registry>/<component>` works with no account and no config. Go to 21st.dev when you want the largest catalogue and you build by pasting into an agent — 12,000+ items, from $6 a month billed yearly. Buy shadcnblocks when you want whole page sections from one hand and a lifetime licence rather than a subscription: $149 once, 2104+ components and 1858+ blocks, single user. Read awesome-shadcn/ui when you want to know why something is on a list at all — it is the only one of the four that publishes its acceptance criteria in full, and the only one running a weekly automated link check across everything it lists. What none of the four does is tell you whether a component's source actually comes back when you ask for it. Three of them publish nothing on the subject, and the fourth — shadcn's own — is admirably direct about why: "community registries are maintained by third-party developers. Always review code on installation." That is the measurement this ranking adds.

Nobody can buy a place on this list. There are no sponsorships, no submissions and no hand-placed entries anywhere on BlockDex. Every number below is recomputed from the nightly crawl, and the formula is printed under the table so you can disagree with it.

eligible624shown25inputs per row4sponsored places0counted2026-09-13
RankRegistryItemsFree shareLast pushManifestScore
1uiableuiable.com829100%5 daysLive site91.52undraw-cnundraw-cn.vaatun.com1,362100%No repoLive site91.03kinetiqkinetiqui.vercel.app1,161100%No repoLive site90.24movingiconsmovingicons.dev555100%3 daysLive site90.05Shadcn UI Blocksshadcnui-blocks.com542100%0 daysLive site89.96animateiconsanimateicons.in542100%8 daysLive site89.97watermelonui.watermelon.sh778100%No repoLive site88.28termcn/uitermcn.dev344100%2 daysLive site88.19@vllnt-uiui.vllnt.com313100%5 daysLive site87.810@react-bitswww.reactbits.dev684100%No repoLive site87.511spectrumuiui.spectrumhq.in272100%0 daysLive site87.312zyeonui.zyeon.ai1,01492%43 daysLive site87.113vlakvlak.dev235100%1 daysLive site86.714@mischiefui.tinkererslabs.com234100%9 daysLive site86.715Animate UIanimate-ui.com580100%No repoLive site86.716vivid-layerwww.vivid-layer.com214100%5 daysLive site86.317ns-uidesign.helpmarq.com542100%No repoLive site86.318lucide-animatedlucide-animated.com467100%22 daysLive site86.219jolyui/uijolyui.dev199100%1 daysLive site86.120keystoneuikeystoneui.io511100%No repoLive site86.021ink-uiink-ui.com168100%1 daysLive site85.422flxui.flexnative.com450100%No repoLive site85.423ruixen-uiruixen.com427100%No repoLive site85.124Sora UIui.soralabs.studio143100%0 daysLive site84.825vault-hyperiuxvault.hyperiux.com142100%2 daysLive site84.8

Inventory 34%, free share 26%, upkeep 26%, manifest served 14%. Every input is on the row, so the ranking can be re-derived. The Manifest column says where registry.json came from, and it deliberately does not wear a verdict stamp: on this site a stamp is the access probe, and where a manifest is served is a different fact about a different request.

The formula, in full

  • Inventory, 34%. Item count, log-scaled. A registry with 4,000 blocks is not forty times more useful than one with 100. Past a few hundred items, more inventory stops deciding whether you find what you need.
  • Free share, 26%. The proportion of items whose source the crawler could actually fetch. A registry that serves no probeable item endpoint scores 40 here rather than 0 or 100, because the answer is not known.
  • Upkeep, 26%. Days since the last push to the registry's public repository. Under two weeks scores full, and over a year scores 14. A registry with no public repository scores nothing here: the axis is dropped and the other three are reweighted to carry the whole score. The row says No repo when that happened.
  • Manifest served, 14%. Whether registry.json comes from the published site (100) or only from the repository (55). A registry you cannot install from by URL is one you have to work to use.

What it deliberately does not measure

Not quality, not design, not accessibility, and not whether the code is any good. A crawler cannot establish any of those. It does not measure popularity beyond the registry's own GitHub stars, which are shown on the registry index and carry no weight here. Stars count how many people bookmarked a repository, not how many shipped with it.

How to use this

Read the top of the table once to know who the serious publishers are. Every item installs by its own URL, so the useful move after that is to search for the component you need across 75,944 indexed items and let the registry be whichever one has it.

Every field in this table is available as JSON at /api/registries. The ranking is computed on this page, so you can apply your own weights to the same numbers.

What is a shadcn registry?

A site or repository that serves a registry.json manifest the shadcn CLI can install from. shadcn's own documentation puts it plainly: "You can use the shadcn CLI to run your own code registry. Running your own registry allows you to distribute your custom components, hooks, pages, config, rules and other files to any project" — and notes that "the registry works with any project type and any framework, and is not limited to React". Community registries listed in the official directory install as `npx shadcn add @<registry>/<component>`, with no extra configuration.

Which shadcn registry should I use?

Use the official directory first — it is free, it is first-party, and the registries in it need no configuration. Use 21st.dev when the catalogue size matters or when your workflow is an agent rather than a terminal, at $6 a month billed yearly. Use shadcnblocks when you want full page sections, a Figma kit and an admin kit, and you would rather pay $149 once than subscribe — remembering that every tier there reads "single user". Use awesome-shadcn/ui to discover what exists rather than to install: it is a curated list whose entries are links, and its own first rule is that a listed resource has to be genuinely about shadcn/ui rather than merely built with it.

Are there free shadcn registries?

Most of the field is free, and the two paid ones are the exceptions rather than the rule. shadcn's own directory costs nothing and requires no account. awesome-shadcn/ui is MIT-licensed, and free use is one of its five acceptance rules — "what does not get in is a product that is paid end to end, with no free tier and no public code". shadcnblocks publishes a free tier behind a login: "basic-tier blocks only require a login", against 22,162 free users and 5,420 paying ones. 21st.dev publishes no free plan on its pricing page at all — three tiers, all paid. BlockDex checks free by asking each registry's item endpoint for the source rather than by reading a pricing page.

How does anything get listed on these?

Two of the four publish a route and two do not. awesome-shadcn/ui publishes the most complete one anywhere in this market: "this page is the full set of rules a submission is judged by - nothing is checked that isn't written here", two ways to submit, one resource per pull request, and five conditions a resource must clear. shadcn's directory has a documented add-a-registry page — "Don't see a registry? Learn how to add it here". 21st.dev publishes no submission route on its pricing or front page, and shadcnblocks publishes none either; what shadcnblocks publishes instead is a first-party production cadence, "50+ new components each month, 100+ new blocks each month", which describes a catalogue written rather than collected.

Does anyone check that the components actually work?

Not in the sense that matters, and only one of the four says so out loud. shadcn's directory states the limit of its own vouching: "Community registries are maintained by third-party developers. Always review code on installation to ensure it meets your security and quality standards." awesome-shadcn/ui runs the only automated ongoing check in the field — a link check that "runs weekly across the whole list and reports broken links in an issue" — but that checks that a URL resolves, not that code installs. 21st.dev and shadcnblocks both answer the question with authorship rather than with a process. BlockDex re-crawls every registry nightly and asks each item endpoint for the source, which is the same request the install command makes.

Every figure in the answers above was read off the page it names and re-checked on or since 2026-09-06. A job re-fetches each of those URLs on a schedule and fails when the wording a figure came from is gone.

Every row above was served by blockdex.thecompound.tech on 2026-09-13, from the crawl of 2026-09-13, and is stored in this repository.

Best shadcn registries (2026) · BlockDex