O seu agente de IA pode executar KYC. Com um comando.
Instale as doze Agent Skills de código aberto da Didit no Cursor, Claude Code, Codex ou OpenCode com um único comando `npx`. Pergunte ao agente em português simples. Ele escreve o código e chama a API em tempo real.
Confiado por mais de 2.000 organizações em todo o mundo.
Porquê skills, e não docs
A documentação é para humanos. As competências são para agentes.
Dê ao Cursor um separador de documentação e ele inventa endpoints e coloca em minúsculas os enums de estado.
Dê-lhe uma skill Didit e cada linha aponta para a API real, endpoint bloqueado,
cabeçalhos bloqueados, scaffold do handler de webhook assinado. Um comando `npx`.
Como funciona
Da instalação npx a um utilizador verificado em quatro passos.
Passo 01
Instale as skills
Execute `npx skills add didit-protocol/skills` no seu projeto. A CLI deteta Cursor, Claude Code, Codex ou OpenCode e coloca cada skill na pasta correta. Instale apenas uma com `--skill didit-face-match`.
Passo 02
Obtenha uma chave API
Ou obtenha uma em 60 segundos em [business.didit.me](https://business.didit.me) e `export DIDIT_API_KEY=…`, ou deixe o agente registar-se programaticamente, `POST /programmatic/register/` seguido de `verify-email/`, e uma chave é devolvida na resposta. Não é necessário browser.
Passo 03
Pergunte em português simples
*“Adicione o KYC da Didit ao meu fluxo de registo.”* *“Verifique este utilizador contra as listas AML.”* O agente lê a skill relevante, escreve o código e configura o handler de webhook assinado. Correto à primeira, sem alternar entre separadores de documentação.
Passo 04
O utilizador verifica
A Didit aloja a câmara, a captura de documentos, a selfie, a transferência móvel. Menos de 30 segundos para o utilizador. Veredito em menos de 2 segundos no seu webhook, assinado com `X-Signature-V2` para que possa confiar.
O que as skills lhe dão
Doze competências. Zero alucinações. Em todas as APIs Didit.
Cada skill é um único ficheiro Markdown que bloqueia o endpoint, os cabeçalhos, o formato do corpo e o enum de estado. Código aberto. Versionado com a API.
Uma competência central para contas, sessões, fluxos de trabalho e faturação (mais de 45 endpoints), além de onze competências autónomas para ID, prova de vida, correspondência facial, pesquisa facial, idade, AML, email, telefone, comprovativo de morada e validação de base de dados.
didit-liveness-detection99.9% accurate selfie liveness
didit-face-matchCompare two faces, 0–100 score
didit-aml-screening1,300+ sanctions + PEP lists
didit-proof-of-addressOCR + geocoding
didit-database-validationGov databases · 18 countries
+ face-search, age, email, phone, 12 in total.
02 · Instalar
Um comando npx. Feito.
`npx skills add didit-protocol/skills` instala todas as doze. Adicione `--skill didit-face-match` para escolher uma. Prefere Git? Faça `git clone` e copie o que precisa. Não quer usar ferramentas? Coloque um único `SKILL.md` na pasta de competências do seu agente.
Detects Cursor, Claude Code, Codex, OpenCode automatically.
03 · Hosts
Cursor, Claude Code, Codex, OpenCode.
A mesma competência Markdown em todo o lado. A CLI deteta automaticamente o seu runtime e escreve para `.cursor/skills/` ou `.claude/skills/`. Construído no padrão aberto [agentskills.io](https://agentskills.io), sem SDK de fornecedor.
Bloqueia o endpoint real, os cabeçalhos reais, o "casing" real do estado, os preços em tempo real. O agente deixa de inventar caminhos e de usar minúsculas para enums. 51 endpoints testados em cada lançamento.
`export DIDIT_API_KEY=…` e a competência está ligada. Ou deixe o agente registar-se via `POST /programmatic/register/` e obtenha uma chave sem sair do terminal.
Todas as competências estão em [github.com/didit-protocol/skills](https://github.com/didit-protocol/skills). Audite o código, personalize os prompts, envie um PR. A competência é lançada no mesmo dia que o endpoint.
Versioned with the API · new endpoint, same-day skill.
Integre
Instale as skills. Pergunte ao agente. Lance.
Esquerda: a instalação. Direita: o cURL que o agente escreve depois de a skill estar implementada.
npx skills addInstalar
$ npx skills add didit-protocol/skills✓ 12 competências instaladas# or just one$ npx skills add didit-protocol/skills \
--skill didit-face-match# set the api key$ export DIDIT_API_KEY="sk_live_..."
OKProntoCursor, Claude Code, Codex, OpenCode detetados automaticamente.
Ou `git clone` o repositório para controlo total.repositório →
Instale todas as 12 Didit Agent Skills com um único prompt.
Cole no Cursor, Claude Code, Codex ou OpenCode. O agente instala as skills, configura o ambiente, estrutura o criador de sessões + o handler de webhook assinado e testa contra a API em tempo real.
didit-integration-prompt.md
You are installing the Didit Agent Skills into this project and wiring the first verification session against the live https://verification.didit.me/v3/ API. After this prompt, the project should pass identity verification — Know Your Customer (KYC), Anti-Money Laundering (AML) screening, document Optical Character Recognition (OCR) — through Didit.
Didit ships twelve open-source Agent Skills at https://github.com/didit-protocol/skills and follows the open standard at https://agentskills.io. The CLI auto-detects whether you are running inside Cursor, Claude Code, Codex, or OpenCode and drops the skills into the right folder.
PRE-REQUISITES
- A Didit API key (DIDIT_API_KEY). Either issued from https://business.didit.me, or self-registered by the agent via POST /programmatic/register/ + POST /programmatic/verify-email/ (no browser needed).
- A workflow_id from the Workflow Builder that bundles ID Verification + Passive Liveness + Face Match + IP Analysis + AML — or just use the didit-kyc-onboarding skill below to create one programmatically.
- A webhook endpoint that verifies the X-Signature-V2 header with HMAC-SHA256 on the raw body bytes (do NOT re-serialise the parsed JSON; the signature will not match).
STEP 1 — Install the skills
Recommended:
npx skills add didit-protocol/skills
This installs all twelve skills. To install only one:
npx skills add didit-protocol/skills --skill didit-face-match
Or git clone the repo and copy what you need:
git clone https://github.com/didit-protocol/skills.git
cp -r skills/didit-kyc-onboarding .claude/skills/
cp -r skills/didit-id-document-verification .claude/skills/
The twelve skills:
- didit-verification-management : the hub. Account, sessions, workflows, billing, blocklist, webhook config. 45+ endpoints
- didit-kyc-onboarding : full KYC recipe (ID + selfie + face match) in one call
- didit-id-document-verification : passports, ID cards, driver's licences. OCR, MRZ, NFC. 4,000+ documents, 220+ countries
- didit-liveness-detection : 99.9%-accurate liveness from a single selfie
- didit-face-match : compare two faces, get a 0–100 score
- didit-face-search : 1:N face search for deduplication and blocklists
- didit-biometric-age-estimation : estimate age from a selfie
- didit-email-verification : email OTP, detects breached / disposable / undeliverable
- didit-phone-verification : phone OTP via SMS, WhatsApp, or Telegram. Catches VoIP
- didit-aml-screening : 1,300+ sanctions, PEP, adverse-media lists. Dual-score risk
- didit-proof-of-address : utility bills, bank statements. OCR + geocoding
- didit-database-validation : government databases across 18 countries
STEP 2 — Set the environment
Every skill reads DIDIT_API_KEY. Session-based skills also expect DIDIT_WORKFLOW_ID. Signed-webhook handlers expect DIDIT_WEBHOOK_SECRET.
export DIDIT_API_KEY="<your api key>"
export DIDIT_WORKFLOW_ID="<your workflow id>" # optional
export DIDIT_WEBHOOK_SECRET="<your secret>" # optional
Programmatic alternative (no browser):
curl -X POST https://apx.didit.me/auth/v2/programmatic/register/ \
-H "Content-Type: application/json" \
-d '{"email": "dev@example.com", "password": "MyStr0ng!Pass"}'
# check the email, get the 6-char code, then:
curl -X POST https://apx.didit.me/auth/v2/programmatic/verify-email/ \
-H "Content-Type: application/json" \
-d '{"email": "dev@example.com", "code": "<code>"}'
# response includes api_key — export it as DIDIT_API_KEY.
STEP 3 — Create a verification session
POST https://verification.didit.me/v3/session/
Headers:
x-api-key: $DIDIT_API_KEY
Content-Type: application/json
Body:
{
"workflow_id": "$DIDIT_WORKFLOW_ID",
"vendor_data": "<your user id, max 256 chars>",
"callback_url": "https://<your-app>/kyc/callback",
"metadata": { "source": "agent-skill" }
}
Response: 201 Created. The hosted session URL is on the `url` field. Redirect the user, or send them the link by email / SMS / WhatsApp. Sub-2-second p99 verdict on completion.
STEP 4 — Read the signed webhook
Didit POSTs to your callback. KYC session statuses are Title Case With Spaces.
Body (excerpted):
{
"session_id": "<uuid>",
"vendor_data": "<your user id>",
"status": "Approved",
"id_verification": { "status": "Approved" },
"liveness": { "status": "Approved" },
"face": { "status": "Approved", "similarity_score": 0.94 },
"aml": { "status": "Approved", "hits": [] }
}
Full enum:
Approved | Declined | In Review | In Progress | Not Started | Abandoned | Expired | Resubmitted | Awaiting User | Not Finished
Verify X-Signature-V2 BEFORE parsing the body — HMAC-SHA256 of the raw bytes with your webhook secret. Re-serialising the parsed body changes whitespace and key order and the signature will not match.
STEP 5 — Read the decision on demand
GET https://verification.didit.me/v3/session/{sessionId}/decision/
Headers:
x-api-key: $DIDIT_API_KEY
Returns the full decision payload — id_verification, liveness, face, ip_analysis, aml. Use this whenever the agent needs to confirm the user's status before allowing an action. Never trust client-supplied "I'm verified" flags.
STEP 6 — Branch on status
Approved → continue
Declined → block, surface decision_reason_code, allow resubmit of the failed step
In Review → wait for the analyst webhook; don't block forever
Resubmitted → user re-took a failed step; new verdict is coming
Awaiting User → user hasn't completed the flow; nudge with a reminder
Expired → create a new session
Abandoned and Declined sessions are NOT billed.
STEP 7 — Optional: ongoing AML monitoring
If AML monitoring is enabled on the workflow ($0.07 per user per year), Didit fires status.updated whenever the user lands on a new sanctions / PEP / adverse-media list. No extra endpoint to call.
WEBHOOK EVENT NAMES
- status.updated : KYC or KYB session status changed
- data.updated : session data corrected after creation
- user.status.updated : User entity changed status (Active, Flagged, Blocked)
- user.data.updated : User entity counters, metadata, or aggregate fields changed
- activity.created : timeline activity recorded
Verify X-Signature-V2 on every payload. The webhook secret is per-environment — sandbox is separate from production.
CONSTRAINTS
- KYC session statuses use Title Case With Spaces (Approved, In Review). Do NOT transform them to UPPER_SNAKE_CASE — that casing is for Know Your Business (KYB) sessions and Transaction Monitoring, not KYC.
- HMAC verification runs against the RAW request body bytes. Never re-serialise the parsed JSON.
- Bundle price is $0.30 (ID + Liveness + Face Match + IP Analysis). AML adds $0.20. 500 verifications free every month, forever.
- Default record retention is unlimited unless you configure it shorter (30 days to 10 years per application).
Read the docs:
- https://docs.didit.me/getting-started/agent-skills
- https://docs.didit.me/sessions-api/create-session
- https://docs.didit.me/sessions-api/retrieve-session
- https://docs.didit.me/integration/webhooks
Skills repo:
- https://github.com/didit-protocol/skills
Start free at https://business.didit.me — sandbox key in 60 seconds, 500 verifications free every month, no credit card.
Precisa de mais contexto? Consulte a documentação completa do módulo.docs.didit.me →
Conformidade desde a conceção
Abra um novo país com um clique. Nós fazemos o trabalho difícil.
Abrimos as subsidiárias locais, garantimos as licenças, realizamos os testes de penetração, obtemos as certificações e alinhamos com cada nova regulamentação. Para lançar verificações num novo país, basta ativar um botão. Mais de 220 países ativos, auditados e testados trimestralmente, o único fornecedor de identidade que um governo de um estado-membro da UE formalmente considerou mais seguro do que a verificação presencial.
Agent Skills de código aberto. Uma por capacidade Didit.
0
Endpoints de API cobertos. Todos testados em cada lançamento.
<0s
Veredito KYC de ponta a ponta por sessão em p99.
0
Verificações gratuitas todos os meses, em todas as contas.
Três níveis, uma tabela de preços
Comece grátis. Pague por utilização. Expanda para Enterprise.
500 verificações gratuitas todos os meses, para sempre. Pague à medida que usa para produção. Contratos personalizados, residência de dados e SLAs (Acordos de Nível de Serviço) no Enterprise.
Grátis
Grátis
$0 / mês. Não é necessário cartão de crédito.
Pacote KYC gratuito (Verificação de ID + Prova de Vida Passiva + Correspondência Facial + Análise de Dispositivo e IP), 500 / mês, todos os meses
Comece grátis → pague apenas quando uma verificação for executada → desbloqueie o Enterprise para um contrato personalizado, SLA ou residência de dados.
FAQ
Perguntas frequentes
What is Didit?
Didit is infrastructure for identity and fraud, the platform we wished existed when we were building products ourselves: open, flexible, and developer-friendly, so it works as a real part of your stack instead of a black box you integrate around.
One API covers verifying people (KYC, know your customer), verifying businesses (KYB, know your business), screening crypto wallets (KYT, know your transaction), and monitoring transactions in real time, on a stack built to be:
Fast, sub-2-second p99 on every session
Reliable, in production with 1,500+ companies across 220+ countries
Secure, SOC 2 Type 1, ISO 27001, GDPR-native, and formally attested by Spain's financial regulator as safer than verifying someone in person
The footprint underneath: 14,000+ document types in 48+ languages, 1,000+ data sources, and 200+ fraud signals on every session. The Didit infrastructure dynamically learns from every session and gets better every day.
What is an Agent Skill?
A small Markdown file that teaches your AI agent how to call one of Didit's APIs, endpoint, headers, body shape, error cases, with working code in TypeScript, Python, and cURL.
Install it once. The next time you ask Cursor or Claude Code “add KYC to my signup flow”, the agent reads the skill and writes the right code on the first try. No tab-switching to docs. No invented endpoints. No wrong status casing.
`didit-kyc-onboarding`, full KYC recipe (ID scan + selfie + face match) in one go
`didit-id-document-verification`, passports, ID cards, driver's licences. OCR, MRZ, NFC. 4,000+ document types, 220+ countries
`didit-liveness-detection`, 99.9% accurate liveness from a single selfie
`didit-face-match`, compare two faces, get a 0–100 score
`didit-face-search`, 1:N face search for deduplication and blocklists
`didit-biometric-age-estimation`, estimate age from a selfie
`didit-email-verification`, OTP, detects breached and disposable addresses
`didit-phone-verification`, SMS, WhatsApp, or Telegram OTP. Catches VoIP
`didit-aml-screening`, 1,300+ sanctions, PEP, and adverse-media lists
`didit-proof-of-address`, utility bills, bank statements, OCR with geocoding
`didit-database-validation`, government databases across 18 countries
Install the lot in one command: npx skills add didit-protocol/skills. Install just one: npx skills add didit-protocol/skills --skill didit-face-match.
How fast is the verification for my end user?
The full flow normally takes under 30 seconds end-to-end, pick up the ID, snap the document, snap the selfie, done. That is the fastest in the market. Legacy KYC providers usually take more than 90 seconds for the same flow.
On the back end, Didit returns the result in under two seconds at p99, measured from the moment the user finishes the selfie to the moment your webhook fires. Mobile capture is tuned for slow phones and slow networks: progressive image compression, lazy software development kit load, and a one-tap hand-off from desktop to phone via QR code if the user starts on web.
Which agents work with the skills?
Cursor, Claude Code, Codex, and OpenCode, plus anything that follows the agentskills.io open standard.
The npx CLI detects your agent and drops the skill into the right place, .claude/skills/, .cursor/skills/, and so on. Same Markdown file everywhere, no host-specific SDK.
Under the hood every skill points at the same https://verification.didit.me/v3/* API with the same x-api-key header and the same signed webhook. The skill is just how the agent reads it.
What happens if a user fails, abandons, or expires?
Every session lands on one of seven clear statuses, so your code always knows what to do:
Approved, every check passed. Move the user forward.
Declined, one or more checks failed. You can allow the user to resubmit the specific failed step (for example, re-take the selfie) without re-running the whole flow.
In Review, flagged for compliance review. Open the case in the console, see every signal, decide approve or decline.
In Progress, user is mid-flow.
Not Started, link sent, user has not opened it yet. Send a reminder if it sits too long.
Abandoned, user opened the link but did not finish in time. Re-engage or expire.
Expired, the session link aged out. Create a new session.
A signed webhook fires on every status change, so your database always stays in sync. Abandoned and declined sessions are free.
Where does my customer data live and how is it protected?
Production data is processed and stored in the European Union by default, on Amazon Web Services. Enterprise contracts can request alternative regions for jurisdictions whose regulators require it.
Encryption everywhere. AES-256 at rest across every database, object store, and backup. Transport Layer Security 1.3 in transit on every API call, webhook, and Business Console session. Biometric data is encrypted under a separate Customer Master Key.
Retention is yours to control. Default retention is indefinite (unlimited) unless you configure shorter, between 30 days and 10 years per application, and you can delete any individual session at any time from the dashboard or the API.
Certifications: SOC 2 Type 1 (Type 2 audit in progress), ISO/IEC 27001:2022, iBeta Level 1 PAD, and a public attestation from Spain''s Tesoro / SEPBLAC / CNMV that Didit''s remote identity verification is safer than verifying someone in person. Full report at /security-compliance.
Is Didit compliant for my industry?
Didit ships compliant by default for the regulators that matter to identity infrastructure:
GDPR + UK GDPR, controller / processor split, full Data Processing Agreement published, lead supervisory authority named (Spain''s AEPD).
AMLD6 + EU AML Single Rulebook, 1,300+ sanctions, politically exposed person, and adverse-media lists screened in real time.
eIDAS 2.0, EU Digital Identity Wallet aligned; reusable-identity ready.
MiCA (Markets in Crypto-Assets), ready for crypto on-ramps, exchanges, and custodians.
DORA, Digital Operational Resilience Act, EU financial-services operational resilience.
BIPA, CUBI, Washington HB 1493, CCPA / CPRA, US biometric privacy (Illinois, Texas, Washington) and California consumer privacy.
UK Online Safety Act, age-gating and child-safety obligations.
FATF Travel Rule, originator and beneficiary data on crypto transfers, IVMS-101 interoperable.
The CLI detects whether you're in Cursor, Claude Code, Codex, or OpenCode and drops every skill into the right folder. Install just one with --skill didit-face-match.
Prefer Git? git clone https://github.com/didit-protocol/skills and copy what you want. Prefer no tooling? Open the repo and copy a single SKILL.md into your agent's skill directory, every skill is self-contained.
The only env var every skill needs is DIDIT_API_KEY. Get one in 60 seconds at business.didit.me, or let the agent register itself programmatically via POST /programmatic/register/.
What does a real prompt look like?
Plain English, no jargon, no boilerplate:
“Register a Didit account for me”
“Create a KYC workflow with ID scan and liveness”
“Create a verification session for this user”
“Screen John Smith against AML databases”
“Check my credit balance”
The agent reads the relevant skill and writes production-ready code in TypeScript, Python, or Bash. No guesswork, the skill ships the exact endpoint, header, body, and status enum.
Why use a skill, not just the docs?
Docs are written for humans. Agents trip on the same things over and over:
Invented endpoints, they make up POST /v3/wallet-screening/ when the real path lives inside /v3/transactions/
Wrong status casing, KYC is Approved, KYB is APPROVED, agents mix them up
Broken webhook signatures, they re-serialise the JSON before verifying the HMAC, which breaks the signature
Stale pricing, they quote old numbers from training data
Skills lock all of that. Versioned alongside the API: when Didit ships a new endpoint, the skill ships the same day. 51 endpoints tested, every release.