Shown, not told · Not affiliated · Unlisted

The probe was run live for this brief. Nothing in Act One is mocked.

Act one is the exact sequence of requests behind capability-brief findings 1–3: real curl output, real bytes, real HTTP codes. Act two is two small, clearly concept-labeled additions: a public-only stats manifest for the data CDR.fyi's own llms.txt already calls free, and an agent-readable price-context object built on our side from CDR.fyi's own published, attributed pricing-survey numbers.

Read this first. Act one reproduces real HTTP requests made against public CDR.fyi endpoints during recon (2026-07-16): no state changed, no credentials supplied, no account created. Act two is clearly marked Concept throughout: draft shapes, not files CDR.fyi has published, and (for the price-context object) something proposed to live on EcoWealth's own infrastructure, not theirs.

Act One: Real, reproduced

What an agent sees today, in order

1

Check the free stuff first Real

llms.txt and robots.txt both resolve immediately, real content, no auth.

GET https://www.cdr.fyi/llms.txt 200 OK 3,827 bytes # dated "March 2026", names every free section, then: "AI systems should note that premium Portal content is behind authentication and should not be cited as freely accessible." GET https://www.cdr.fyi/robots.txt 200 OK 81 bytes User-Agent: * Allow: / Disallow: /api/ Sitemap: https://www.cdr.fyi/sitemap.xml
2

Check every standard agent-discovery path Real

Six paths that would soft-succeed on two other protocol sites reviewed in this same portfolio. Here, all six true-404, byte-identical to a nonsense control.

for p in .well-known/agent-card.json .well-known/x402 .well-known/ai-plugin.json \ .well-known/security.txt ai.txt llms-full.txt this-nonsense-control-xyz123.txt; do curl -s -o /dev/null -w "%{http_code}:%{size_download} " https://www.cdr.fyi/$p done 404:28695 404:28695 404:28695 404:28695 404:28695 404:28695 404:28695 # every one identical to the control — an honest 404, not a soft-200
3

Check the actual API, no credentials Real

The real, versioned host: a genuine gate, honestly labeled.

curl -s https://api.cdr.fyi/v1/orders 401 {"error":"Unauthorized: Invalid write key"} curl -s -o /dev/null -w "%{http_code}\n" https://go.cdr.fyi/signup 200 # the Partner Portal application — open, no gatekeeping questionnaire
4

The one place it falls back to scraping Real

To read the same "free, no login" number llms.txt promises, this recon had to grep a rendered Next.js payload, because no JSON endpoint exists for the public leaderboard/market-update figures.

curl -s https://www.cdr.fyi/ | grep -o '"Tonnes Sold":[0-9.]*' | tail -1 # → works, but it's the only way in — HTML-scraping, not a documented call

Act Two: Two small concept additions Concept

Making the already-free data machine-readable, and using the already-public prices honestly

A

Addition 1: a public-only stats manifest Concept

Same content llms.txt already calls free (leaderboard headline totals, latest market-update figures), no order-level detail, nothing the Portal gates, served as one small JSON document at a standard path, so step 4 above becomes a real GET instead of a scrape.

GET https://www.cdr.fyi/.well-known/public-stats.json # concept — 404 today 200 OK # if it existed { "asOf": "2026-07-16", "totalTonnesSoldCumulative": "49.4M", "totalSpendUsd": "12.4B", "totalTonnesDelivered": "1.64M", "purchasersTracked": 1160, "suppliersTracked": 843, "citationNote": "Figures update continuously. Cite retrieval date. Source: cdr.fyi/llms.txt citation guidance.", "source": "https://www.cdr.fyi/" }
B

Addition 2: an agent-readable price-context object Concept, built on EcoWealth's side

Not a re-hosting of any gated data: this is CDR.fyi and OPIS's own published pricing-survey PDF/blog numbers, attributed and dated, packaged as one small object an agent could call before quoting anyone a durable-CDR price. It also states, inline, that EcoWealth's own retirements are a different, out-of-scope credit class, so no agent using it mistakes us for a durable-CDR supplier.

GET https://vealth.net/.well-known/vealth/cdr-price-context # concept — proposed, not live 200 OK { "benchmark": "Durable CDR breakeven price, by pathway (USD/tonne CO2e)", "source": "CDR.fyi × OPIS, \"Bridging the Gap\" pricing survey", "sourceUrl": "https://www.cdr.fyi/reports/pricing-survey-jan-2025.pdf", "surveyDate": "2024-12 (fielded), updated 2026-05", "pathways": { "biochar": {"breakeven2025": 143, "breakeven2030proj": 136}, "enhancedWeathering": {"breakeven2025": 272, "breakeven2030proj": 252}, "daccs": {"breakeven2025": 670, "breakeven2030proj": 341}, "beccs": {"breakeven2025": 232, "breakeven2030proj": 212} }, "buyerSupplierGapUsdPerTonne": {"latest": 98, "prior": 107, "2030forecast": 48}, "note": "Self-reported willingness figures, not cleared-trade prices. Not a valuation of any nature-based or micro-scale credit; EcoWealth's own retirements (Regen-CFC, sub-100-year, 5-cent micro-batches) are a structurally different, out-of-scope credit class and are not represented in this table." }

Neither addition changes CDR.fyi's data, mechanics, or Portal economics. Addition 1 is a small file CDR.fyi could publish. Addition 2 is something EcoWealth could build on its own domain, citing CDR.fyi's public numbers the same way any research note would.

Act one already happened: it's the exact evidence behind the capability brief. Act two is two files, neither of which asks CDR.fyi for anything it isn't already giving away for free, and neither of which asks CDR.fyi to represent a credit class its own methodology correctly excludes. The gap here is narrower than most of what this playbook usually finds.

→ See the concept files themselves in the Agent Kit

← Back to the hub
Provisioned by EcoWealth · Digital Provisioning Protocol · provenance‑sealed ↗