Checks reference
🔑 Auth: SDK client (App API key) · 💾 Run in a workflow with the user’s
valyd_access_tokenand the passed proof saves to their Valyd ID
Every check Valyd can run, in one place: what it verifies, what the user provides, what comes back, and where it’s available — inside a Reusable Verification workflow (a connected user’s session, where every check runs) and/or as a direct Unique Human API call (API key only, no user login).
ID verification — id_verification
Verifies a government ID: OCR of the document fields plus an authenticity score. The user provides the front (and optionally back) image of the ID. You get back a proof that the document was verified; the raw extracted fields stay encrypted on the account. Use it whenever you need to know who a document says someone is. Available: Reusable Verification workflow
Liveness — liveness
Passive liveness on a single selfie: is this a real, live capture? Use it as the cheap first gate before a face match. Available: Reusable Verification workflow
Anti-spoof — antispoof
Stronger “is this a live human capture?” answer with a vendor-neutral human_score (0–100). The
user provides a single image (score capped at 85) or a 3–8 frame burst captured over ~2s, which
adds motion and same-person consistency analysis. A workflow session captures a live camera burst
with a random on-screen action for the strongest assurance (assurance: "captured").
Use it when presentation attacks (photos of photos, replays) are a real threat.
Available: Unique Human API — Liveness (no-account workflow session) · Reusable Verification workflow
Anti-spoof + identity — antispoof/identity
Runs the identical anti-spoof pipeline and, only if it passes, resolves the proven-live face to a
stable valyd_ uuid from the global face gallery. You get the human_score plus
identity: { valyd_uuid, registered } — the same face resolves to the same uuid whenever the
gallery match clears its similarity threshold. Use it for duplicate-account / sybil detection with
liveness assurance built in.
Available: Unique Human API (no-account workflow session) · Reusable Verification workflow
Face uniqueness — face-uniqueness
One face = one Valyd uuid. Enrolls or matches a selfie against the global gallery and returns the
stable valyd_uuid plus whether it was newly registered. Use it to stop one person opening many
accounts.
Available: Unique Human API — Uniqueness (no-account workflow session) · Reusable Verification workflow
Face match — face_match
1:1 comparison of two face images — typically the ID portrait against a fresh selfie. You get
similarity and the pass threshold (default ~0.95). Use it to bind a live person to a verified
document.
Available: Reusable Verification workflow
Age verification — age
Computes age bands from a verified date of birth. With a valyd_access_token the bands are
computed from the account’s KYC-verified DOB and returned as a proof. Use it for age-gated products.
Available: Reusable Verification workflow
License / credential verification — credential
Looks up a professional license in the provider registry and matches it to a name. In a workflow
session the name comes from the verified ID (never client-supplied); you get match and the
registry’s license record (status, expiry, specialty). Registry lookups can take 10–60 s. Use it
to verify doctors, nurses, and other licensed professionals.
Available: Reusable Verification workflow
KYC + credential — combined
ID verification + liveness + face match + license lookup in one workflow session. The license is
matched against the name OCR’d from the ID — never a client-supplied name — so a caller cannot
substitute an arbitrary name to claim someone else’s license. You get a per-check breakdown plus
the verified identity proof. status is "passed" only when every check passes.
Available: Reusable Verification workflow
Location — location
Records and validates a geolocation fix for a session. Used by workflows like EVV (electronic visit verification) to prove where a check happened. Available: Reusable Verification workflow
Biometrics are irreversible vectors, never images. Valyd does not store or return face images. Enrollment converts a selfie into a one-way biometric vector (template); every later face match compares vectors. The photos you submit to a check are processed transiently for that check and are not retrievable from a Valyd account. The template is never exposed through any API, and the ID
portraitin a KYC result is extracted from the document you submitted in that request — not a stored account photo. Full scoping →
Next: bundle checks into a workflow, or — for the direct liveness and uniqueness calls — see the Unique Human API.