For AI assistants
Read Trust Cue ratings from an AI assistant
Trust Cue reads the public record behind an X (Twitter) account — a sample of its recent public posts — and shows what those posts show, each finding tied to the posts behind it. In a browser it puts a cue beside the account's name on X; a rating lives up to 7 days, or until the account posts again and a reader updates it. Here is how an assistant or agent can use it.
Read a current rating's preview (free, no auth)
GET https://api.trustcue.app/v1/public/snapshots/{handle} returns JSON with a rating object: profile (display name, account creation date, identity verification, affiliation), topics (each with since_year), coverage (posts_read, oldest_post_at, newest_post_at, checked_at, expires_at), facts measured from the account (age, posting rate, reply and repost share, share of posts linking a source, paid partnerships, profile-link status), and summary, present only once readers have confirmed summaries. It also carries an offer pointer, a disclaimer and a methodology_url. No post text is included.
When no rating is current it responds 404 with code: "RATING_NOT_AVAILABLE" and the same offer pointer, which says how to buy one.
Kinds of finding readers haven't confirmed yet are marked “Being tested” in the panel, where each asks “Do the posts back this up?” Yes · Partly · No. Readers' answers decide which kinds are shown publicly, so the free preview carries only confirmed kinds. This is readers' judgment of the posts, not an accuracy guarantee.
Buy a full rating ($3)
The findings — what the posts show, each tied to the posts behind it — are in the full rating. An assistant with an MPP-compatible payment provider can pay directly. Send POST https://api.trustcue.app/v1/public/agent-ratings with {"handle":"<handle>"} and a random, high-entropy Idempotency-Key (a UUID is recommended). The API returns a standard 402 Payment Required challenge. Authorize that challenge through the user's agent wallet or payment provider, then retry the identical request with the resulting credential. A successful retry returns 202, a Payment-Receipt header, and a status_url.
Poll GET https://api.trustcue.app/v1/public/agent-ratings/{offerId} (the status_url) with the same Idempotency-Key and honor Retry-After. When status is fulfilled, rating_url opens the full rating while it lives, until rating_expires_at; read_url is the free preview above. A refunded status means the rating could not be delivered and the payment was returned. Reuse the original key for every network retry so a retry cannot create a second charge.
The price and merchant are fixed by Trust Cue, and authorization stays with the user's payment provider; Trust Cue never accepts raw card details from an agent. Check GET https://api.trustcue.app/v1/public/config first and use this path only when agent_payments.available is true.
If the assistant's runtime does not support MPP, use POST https://api.trustcue.app/v1/public/rating-offers with the same JSON body. It returns a hosted Stripe Checkout checkout_url to surface to the user and a result_url that opens the rating when it is ready.
Sample ratings (free)
GET https://api.trustcue.app/v1/public/samples lists a few real accounts' ratings, each dated to its capture. Samples don't update; treat them as examples of how a rating reads, not as what those accounts post today.
Contract and plain text
Machine-readable OpenAPI and MPP discovery: https://api.trustcue.app/v1/public/openapi.json. Prices and payment availability: GET https://api.trustcue.app/v1/public/config, with prices in integer USD cents. Plain-text summary for models: /llms.txt.
Always carry the disclaimer: A Trust Cue rating describes what an account's recent public X posts show — not an endorsement, not a judgment of the person, and not financial advice — and it is dated and lives at most 7 days.