Reviewed crypto card data for software and people

API / Agents

cryptocard.guide turns reviewed provider evidence into comparable card profiles and explainable rankings. The same research is available through normal web pages, JSON, Markdown and public recommendation endpoints, so an AI agent can find a suitable card, explain the result and cite the evidence behind it.

Reviewed facts, ready for questions

A visitor can browse the catalogue and compare cards visually. An agent can ask the same research layer a structured question without scraping tables or guessing what a field means.

Example question

Find crypto cards for a US user that accept USDC funding, then rank them for stablecoin spending.

The API treats US as country context, applies the funding and asset requirements, ranks the remaining cards with the requested scenario and returns reasons, unknowns, sources and any relevant current offer.

What you can use

How agents find the API

The root /llms.txt file points to this guide, the card index, recommendation endpoint, offers endpoint and OpenAPI document. An agent that starts on a card or catalogue page can follow the same links from there.

These endpoints are read-only and do not require an account or API key. Responses include links back to the equivalent browser result and the canonical pages suitable for citation.

Card data formats

/api/v1/cards returns the published card index. /api/v1/cards/{slug} returns one reviewed public profile as JSON. Send Accept: text/markdown to either card endpoint for a compact Markdown representation.

Card records include the canonical page, last-checked date, source confidence, selected evidence, explicit unknowns, score explanation, structured funding routes and the current application link. Machine representations are marked noindex. The normal HTML page remains the canonical citation and search URL.

Use a web link or the matching API

The web pages and APIs use the same public parameters. A user can open the result in a browser, while an agent can request the JSON form of the same question.

The recommendation API also supports an additional agent-focused ranking lens: /api/v1/recommendations?country=US&funding=crypto&asset=USDC&scenario=stablecoin_spend. The stablecoin_spend scenario is not shown as a main catalogue tab.

Recommendation parameters

country
US, CA, GB, DE and other country codes available in the site selector. EU, EEA and ALL are also accepted.
funding
any, crypto, bank, card or internal. Crypto includes reviewed on-chain funding from an external wallet. Internal means a transfer within the provider's own account or card system. The API does not connect to or inspect a wallet.
asset
One asset code, for example USDT, or a comma-separated list, for example USDC,USDT.
scenario
overall, cashback, travel, funds_control, low_fees, stablecoin_spend or transfers.
limit
From 1 to 50 results.

Funding and asset requirements are strict. Unknown support does not pass them. Country is soft context: a card is excluded only by explicit unavailability or a complete allowlist that omits the selected country.

Why a card fits, then what the user can get

Each result first explains the product fit: score reasons, country context, funding routes and unresolved unknowns. It then adds the current application link and any relevant offer. This gives an agent enough information to explain the card before mentioning a promotion. In JSON, these groups are named fit and commercial.

Offers do not change card scores, ranking order or recommendation eligibility.

Offer value and country context

Each offer describes what an eligible user may receive, the requirements, deadline and whether its value can be compared with another offer. A return-at-requirement percentage is calculated only when a fixed benefit and minimum spend use the same asset. Up-to rewards, multipliers and unlike currencies remain explicitly non-comparable.

Card availability and offer eligibility are reported separately. If offer terms do not confirm the selected country, the response says so instead of assuming eligibility.

Citation guidance

API responses provide the useful machine-readable context, while the linked canonical page gives a stable public citation target.

What makes an answer trustworthy

MCP compatibility

MCP is not required to use the data today. If agent clients need tool-style access later, a future MCP server can wrap the same card search, recommendations, offers and source lookup. It would use the same reviewed facts and ranking logic as the web and API.