Open data
Building the Try tools meant compiling reference data that was scattered, undated or uncited. All four compilations are published openly, with the source and the verification date attached to every figure, under CC BY 4.0.
Each is generated from the data the product itself runs on, so a dataset cannot quietly drift from the tool it documents. Nothing here is scraped or auto-updated — a person reads every change, because a confidently wrong figure is worse than an old one.
250 photo and signature requirements for passports, visas and exams. 124 link to the government or embassy page they came from; every record carries the date it was last checked by hand. Records without a citable public source are published and labelled as presets rather than hidden.
Seven sets across five jurisdictions — US and Canada federal income tax, India income tax and GST, UK VAT, US CPI, Australian HECS. Each names its issuing authority, the exact source page, the date a human verified it and the date that verification expires.
Which published document defines each tool's behaviour — RFCs, NIST FIPS, IANA registries — with a link to the spec. Derived from the tool pages at build time, so it cannot claim a citation a tool no longer carries.
Published list prices for OpenAI and Anthropic models in USD per million tokens, pinned to a verification date and checked in CI against OpenRouter for drift. List rates for planning — not invoices, which differ with caching, batch tiers and negotiated pricing.
Reuse any of them with attribution. If you cite one, how to cite Acsaven has the preferred wording.