🔐 eKYC Suite

Financial-grade identity verification for AI agents. 8 capabilities in one package.

---

What it does

Upload a photo or video, ask a question, get structured verification results. That's it.

#	Capability	What you get
1	Face Comparison	Similarity score 0-100 between two face photos
2	Photo Liveness	Is this photo real or AI-generated/deepfake? (risk level 1-3)
3	Video Liveness	Is this video real or deepfake? (risk level 1-3)
4	ID Card OCR	Name, ID number, address from Chinese national ID card
5	Bank Card OCR	Card number and expiry date
6	Driver's License OCR	License number, vehicle class, validity dates
7	Vehicle License OCR	Plate number, VIN, owner, engine number
8	Media Labeling	15+ risk attributes: mask, hat, coercion, tattoo, phone, in-car, etc.

Setup

1. Get API credentials

• Key A (KYC_APPID + KYC_SECRET) → enables capabilities 1-7
• Key B (LABEL_APPID + LABEL_SECRET) → enables capability 8

How to get: Tencent Cloud Console (Key A) or contact Huiyan tech support (WeChat: blue-201809) for Key A + Key B.

⚠️ Use TEST credentials (free 100 calls per appid). Production IDs incur charges.

2. Configure

Copy .env.example to .env and fill in your test credentials.

How to Use

Install this Skill in your AI agent (OpenClaw, etc.), then just talk to it naturally:

You say	What happens
"Compare these two photos — same person?" + upload 2 photos	Returns similarity score (e.g. 96.2/100)
"Is this photo AI-generated?" + upload a selfie	Returns risk level 1-3 with attack type classification
"Is this video real or deepfake?" + upload a video	Detects deepfake/replay, returns risk level
"Read this ID card" + upload ID card photo	Extracts name, ID number, address, ethnicity, birth date
"Read this bank card number" + upload bank card photo	Extracts card number and expiry date
"Read this driver's license" + upload photo	Extracts license number, name, vehicle class, validity dates
"Read this vehicle license" + upload photo	Extracts plate number, VIN, owner, engine number
"Check for hat and mask" + upload photo	Detects 15+ risk attributes (see table below)

The AI agent handles everything automatically — you just upload files and ask questions in natural language.

Media Labeling — Risk Attributes (Capability 8)

Portrait Attributes — Detect user status and fraud risk

Code	Attribute	Use case
A09	Under coercion	Anti-fraud: detect if the person is being coerced during identity verification
A10	Unconscious / asleep	Risk alert: detect if the person's eyes are closed or eyelids are being forced open
A15	Critical patient	Loan fraud prevention: flag verification attempts by critically ill individuals
A11	On the phone	Third-party guidance detection: someone may be coaching the user through the process
A04	Wearing headphones	Third-party guidance detection: hidden earpiece coaching during verification
A02	Medical mask	Obstruction: face partially covered, may compromise identity verification accuracy
A14	Wearing hat	Obstruction/disguise: may indicate attempt to alter appearance
A06	Sunglasses	Obstruction: eyes hidden, reduces face recognition confidence
A01	Facial sheet mask	Obstruction: skincare mask covering facial features
A05	Nudity	Compliance: flag inappropriate content during video verification
A13	Tattoo	Risk profiling: visible tattoo detected for feature marking

Environment Attributes — Detect business scenario

Code	Attribute	Use case
B02	Multiple people	Fraud risk: third party present during what should be a solo verification
B03	Inside a car	Auto loan verification: confirm the applicant is at/in the vehicle
B06	Hotel room	Risk control: flag unusual verification location for loan/account applications
B07	Car dealership	Consumer finance compliance: verify the applicant is at the dealership

Max 5 label codes per request. Example: "A09,A10,A02,B02,B03"

License

MIT — Copyright (c) 2024-2026 Wefi AI Team