Our Project · Reference Encyclopedia
Scan a payment QR in Zagreb and one in Manila and you're looking at two unrelated standards. Barcoder catalogues all of them – 463 visual-code standards across 197 countries, cross-referenced with the 1,076 wallets that read them.
Answers are scattered across central banks, ISO/IEC specs, vendor PDFs and regulators, in a dozen languages. Barcoder cross-references them.
Payments engineers integrating a new market. Fintech teams sizing wallet coverage before launch. Wallet providers checking which rails they can ride. Researchers who need a citation, not a forum post.
Every page names its sources. Vol. I, Edition 2026 – and it grows with each commit.
Every entry links to the others. Start from a standard, a country, or a wallet – you end up in the same graph.
01
Every QR, barcode and visual-code specification we can find evidence for – EMV merchant-presented modes, national instant-payment QRs, retail GS1 codes, niche transit codes.
02
A page per country covering which schemes operate there, what visual codes those schemes use, and how interoperability works at the merchant counter.
03
One page per wallet – GCash, Alipay, Revolut, M-Pesa and the rest – covering which standards each supports, in which markets, and against which underlying rails.
A reference book you can run. Scan a code, paste a payload, or generate one – and land on the standard that explains it.
Plenty of tools read a QR code. Barcoder tells you what it is. Decoded payloads are matched against all 463 standards and 1,076 wallets using URI schemes, TLV structures (EMVCo MPM), contact formats and payment-scheme identifiers.
Point your phone at a code in the wild. Decode and identify in the browser – no app to install.
Drop in a raw payload string. Get the matching standard and every field broken out and labelled.
112 real codes from schemes around the world. Test your own parser against something authentic.
40 encoders. Fill in the form, generate a valid code, then scan it back – all on one page.
Every Barcoder page renders from one markdown file in the repo. No CMS, no admin backend, no production-only edits.
Each country starts as long-form research – central bank sources, scheme docs, regulator filings, read and written up.
Claude Code skills pull structured per-standard and per-country pages out of that research, consistently every time.
Pages land as markdown in the repo. Every published claim is traceable to the commit that introduced it.
The backend reads those files, the frontend renders them. Boring on purpose – and reproducible from source.
A reference work is only worth as much as its sourcing. Because the content is the repo, any page can be traced back through git to the change that wrote it – and corrected the same way anyone corrects code.
197 country reports is a research-team-year of work. AI agents did the extraction and drafting; a human set the structure, checked the sources and shipped it. That's the whole LumiVerse thesis in one project.
Barcoder is what happens when you point AI research pipelines at a scattered field and make every output auditable. The same approach works on your industry's mess.