Built from live public pages this session: the Puro Registry's own counters, its methodology scope, and its market-position claims, plus one fresh agent-discovery probe. Every claim below cites its source and a re-runnable command.
No jargon. Puro's registry and methodology library are real and credited below; the one fresh gap is worth naming plainly. Here's what closing it is worth.
Every agent-discovery path is a true 404 today, so an assistant asked about Puro's CORCs or methodologies has no machine-legible surface and paraphrases from scraped marketing instead of your live 1.78M-credit ledger.
The Puro Registry's own homepage counters show 1,776,510 credits issued and 1,040,750 retired across 117 certified projects, browsable by methodology, facility, auditor, owner, and location. This is a real, operating ledger, not a marketing number, the same registry Exomad Green's and Biomass Projects' eventual credits would land in.
Puro.earth's own site lists eight certified removal methodologies spanning durability classes from 200 years (biochar, microalgae) to 1,000+ years (DAC+ocean storage, geologic storage, enhanced rock weathering, carbonated materials). The Biochar Methodology alone (Edition 2025, ~142 pages, approved June 2025) covers the entire supply chain from biomass sourcing through application, not just the pyrolysis step.
Puro both writes the biochar/CDR methodology rules and operates the CORC-issuing registry, a dual role. Our earlier public research on Carbonfuture (2026-07-16) found Puro's own quality lead calling supplier-data handling "95% of the workload," which is exactly the pain independent dMRV platforms like Carbonfuture were built to absorb so the standard body doesn't handle raw supplier data itself. Puro's methodology requires calibrated weighbridges, accredited-lab Corg/H-ratio and heavy-metal analysis, and VVB sign-off before a CORC issues, a different order of rigor from EWP's photo+GPS+signature proof, which proves task completion, not carbon quantity or durability.
puro.earth ships a real, working robots.txt (200, Yoast-generated, allow-all, points at a real sitemap). But llms.txt, llms-full.txt, ai.txt, .well-known/agent-card.json, and openapi.json all return a TRUE HTTP 404, a 121,446-byte WordPress error page, byte-identical to a random nonsense-path control fetched in the same pass. That's the honest outcome: a real 404, not a soft-200 fake, but zero agent-legible surface today for any AI representing or evaluating Puro.
