Shopstr and the strange economics of censorship-resistant commerce
- The farmer and the marketplace
- How the thing works
- The wedge
- We have seen this movie before
- The zapsnag and the AI agent
- No safety net
- What this actually is
The farmer and the marketplace
Raw milk is legal to sell on-farm in Pennsylvania. A farmer there can get her herd tested, pass state inspection, and sell directly to anyone who shows up. What she cannot do is list her product on Amazon, eBay, or Facebook Marketplace. Their content policies prohibit it. Not because what she is doing is illegal where she is, but because it is illegal somewhere, and platforms that operate everywhere default to the strictest set of rules.
So she sells at a farmstand. She sells to people who already know about her. The person in Philadelphia who would drive two hours for unpasteurized dairy never finds out she exists.
This is the kind of problem #Shopstr was built for. Not the Silk Road kind. The boring kind, where a legal product meets a content policy written for a different jurisdiction and the seller just vanishes.
How the thing works
Shopstr is a #nostr marketplace client. A seller creates a product listing, which is published as a Kind 30402 event (NIP-99, classified listings) to whatever relays the seller uses. The listing has a title, description, price in sats, images uploaded to Blossom servers, and category tags. Because the listing is a standard Nostr event, it is signed by the seller’s keypair. Anyone running a client that reads NIP-99 events can see it.
Payments are #bitcoin only. Lightning invoices for instant settlement. Cashu ecash tokens for privacy. Nostr Wallet Connect for people running their own Lightning nodes. Shopstr removed fiat payments entirely in February 2026. If you do not hold bitcoin, you cannot buy anything.
Buyer-seller communication runs over NIP-17 gift-wrapped messages. This is the same encrypted messaging protocol I wrote about in my NIP-17 article. The relay sees encrypted blobs. It does not know what you ordered, where it ships, or how much you paid. Even the relay operator, if they wanted to, could not reconstruct a transaction from the data passing through their server.
There are no listing fees. No transaction fees. No subscription. Shopstr charges zero percent. The company behind it, Shopstr Market Inc., presumably runs on something, but the revenue model is not public. The marketplace itself is free in the way that a protocol is free: nobody takes a cut because there is nobody in the middle to take one.
Compare that to eBay, which charges 13.25% of the final sale price. Or Etsy at 6.5% plus a listing fee. Or Amazon, which takes between 6% and 45% depending on the category. Shopstr’s pitch is not just censorship resistance. It is also that the rent is zero.
The wedge
Every censorship-resistant marketplace in history has faced the same pattern. The first products that show up are the ones that cannot be sold anywhere else.
Silk Road launched in 2011. Its primary inventory was illegal drugs. The marketplace processed an estimated $1.2 billion in transactions over two years. It had other categories, but drugs were the engine. The platform was shut down, its founder arrested, and the name became shorthand for “this is what happens when you build a marketplace with no rules.”
OpenBazaar launched in 2014 as the legitimate alternative. A decentralized marketplace with no fees, no accounts, and no content moderation. The vision was eBay without a company in the middle. It struggled for seven years to attract mainstream sellers. The people who needed censorship resistance were not the people selling vintage clothing and handmade candles. OpenBazaar shut down in 2021.
Bisq has survived since 2014 by not fighting the pattern. It is a no-KYC #bitcoin exchange. That is what it does, and that is what its users need. Bisq never tried to be a general marketplace. It found the one niche where censorship resistance is the primary value proposition and stayed there.
The logic is simple. If you sell legal goods on a legal platform, you already have buyers, trust systems, and payment processing. Switching to a censorship-resistant marketplace costs you all of that and gives you a smaller audience. The only sellers with a strong reason to switch are the ones who have been kicked off the platforms they were using or cannot list their products at all.
This creates a selection effect. The early catalog of any permissionless marketplace skews toward goods that mainstream platforms will not carry. Raw milk. Unregulated supplements. Homemade food without commercial kitchen certification. Firearms parts in jurisdictions where they are legal but platform-banned. Seeds for plants that are legal in some states and not others.
None of these are inherently criminal. Most of them are legal in at least some jurisdictions. But they share a common trait: someone, somewhere, decided they should not be listed.
We have seen this movie before
The Silk Road comparison is the one people reach for, but it is the wrong one. Silk Road was designed from the start for illegal transactions. Tor routing, Bitcoin payments, a tumbler built into the platform. The architecture was optimized for hiding.
Shopstr is the opposite. Listings are public Nostr events on public relays. Anyone can read them. The encryption covers the transaction details, not the catalog. You can browse Shopstr without logging in. The product listings are as visible as any Nostr note.
The better comparison is Craigslist in 2005. Craigslist had no identity verification, no payment processing, no escrow, and no real content moderation. Buyers and sellers met in parking lots and exchanged cash. The legal activity was real. The illegal activity was also real. And Craigslist survived because the legal use cases generated enough volume and enough value to outweigh the problems.
But Craigslist also had something Shopstr does not: a massive existing user base. Craigslist could afford a rough product because millions of people were already there. Shopstr has the rough product without the millions. The entire Nostr network has about 17,000 daily active users. The subset of those who want to buy or sell physical goods on a Bitcoin-only marketplace is some fraction of that already small number.
The zapsnag and the AI agent
Two features in Shopstr do not have equivalents anywhere else, and they are worth understanding even if the platform stays small.
A zapsnag turns any Nostr note into a flash sale. A seller writes a regular Kind 1 note, includes the hashtag #zapsnag, a price in sats, and optionally an image. Shopstr’s client parses the note, renders a buy button, and handles the payment via a Lightning zap to the seller’s address. The buyer enters shipping information, which gets encrypted and sent as a gift-wrapped message. Inventory is tracked by counting zap receipts on the original note. When the zap count reaches the quantity, the item shows as sold out.
A text post becomes a storefront. No product listing page. No checkout flow. Just a note that says “handmade soap, 5000 sats, ships anywhere” and a button. You click, you pay, the seller gets your address via encrypted message. That is the entire transaction.
The second feature is newer. In March 2026, Shopstr added an MCP server that lets AI agents browse products, search the catalog, and place orders programmatically. The agent authenticates with an API key, pays via #lightning or Cashu, and the order ships to whatever address the agent provides. Shopstr published an agent.json manifest following the Open Agents schema so that AI systems can discover and interact with the marketplace without human intervention.
I do not know whether AI agent commerce will become a real thing. But if it does, the first marketplace to natively support it is a zero-fee Nostr marketplace where the agent does not need to create an account, verify an identity, or agree to terms of service. Make of that what you will.
No safety net
Here is where I push back on the whole thing, because the problems are real.
Shopstr has no escrow. You pay a seller, they do not ship. Now what? You send them an encrypted message asking nicely. That is it. NIP-85 reviews exist and they are portable across Nostr clients, which is genuinely useful, but when the user base is small a few fake reviews can dominate.
Fraud is cheap. A seller can list a product, collect Lightning payments, and disappear. Create a new keypair, do it again. The Web of Trust filtering that makes Nostr’s social layer workable has not crossed over to commerce in any meaningful way yet.
Then there is the addressable market problem. Most people do not hold bitcoin. Most people who hold bitcoin do not use Lightning. Stack those filters: must run a Nostr client, must pay in bitcoin, must trust a pseudonymous seller with no buyer protection. Each one eliminates most potential customers. You are left with a very committed few.
Shipping is on the merchant. No tracking integration. No delivery confirmation. The seller says it shipped and you believe them or you do not.
The legal picture is untested. Shopstr Market Inc. is a real incorporated company, but no court has addressed what it means to run a permissionless marketplace where encrypted messages prevent even the platform from seeing transaction details.
What this actually is
Shopstr is not going to replace eBay. The user base is too small, the payment methods too narrow, the trust mechanisms too immature. But if you’ve read my article on the mass exodus that keeps not happening, you might remember Shopstr as one line in a list of things being built on Nostr. What I did not say there is that it might be the clearest test case for whether the protocol works beyond social messaging. A marketplace has to deliver trust, payment, and physical goods. That is harder than delivering text.
The raw milk farmer in Pennsylvania does not care about any of this. She wants to list her product somewhere that will not take it down. Shopstr lets her do that today. Whether anyone finds it there is a different problem, and it is the one that censorship-resistant commerce has never solved.
#nostr #bitcoin #lightning #decentralization
Write a comment