AI-Agent First

Peppol for
AI Agents.

The Peppol API designed to be fully bootstrappable by AI. Your coding agent can create an account, send test invoices, and go live. All without human intervention. Minutes, not months.

Start building →Read the docs

Works with your AI agent

Build Peppol integrations with AI

Whether you use Cursor, Claude Code, Codex, or any other AI coding agent, peppol.sh is designed to be discovered, understood, and integrated autonomously.

Why AI-agent first

APIs should be bootstrappable by code

Most APIs require manual signup, dashboard configuration, and email verification. Not peppol.sh. Every step is API-driven.

Create account via API

POST your email to /v1/signup. Get a sandbox API key back instantly. No email verification required to start testing.

Send test invoices immediately

Use your ps_test_ key to send invoices to sandbox. Test the full flow before going live.

Go live when ready

Once KYC gets approved, get your live key. Your AI agent handles the entire onboarding.

AI agent bootstrap

AI Agent

└→ POST /v1/signup

└→ API key returned

└→ POST /v1/invoices

└→ Test invoice sent

└→ POST /v1/account/kyc

└→ POST /v1/account/keys

status: live ✓

What makes it AI-ready

Designed for autonomous integration

Every design decision prioritizes discoverability and automation. AI agents can explore the API without reading documentation.

API-first bootstrapping

Create accounts, manage API keys, submit KYC, and go live. All via API. No dashboard clicks required.

SKILL.md included

We include a SKILL.md file that AI agents can discover and use to send invoices and track delivery autonomously.

OpenAPI spec

Complete OpenAPI 3.1 specification. AI agents can introspect endpoints, parameters, and response schemas.

Predictable errors

Structured error responses with machine-readable codes. AI agents can understand and recover from errors automatically.

Sandbox from signup

Every account starts in sandbox mode. AI agents can experiment without risk before going live.

JSON in, Peppol out

No UBL XML knowledge required. Send simple JSON, we handle the Peppol compliance. AI agents understand JSON natively.

OpenAPI 3.1 spec

SKILL.md

API-first bootstrap

Structured errors

Instant sandbox

SKILL.md

Built on the open standard

Our SKILL.md file lets AI agents discover and use peppol.sh capabilities automatically. Send invoices, check delivery status, and more.

send_invoice

Send a Peppol invoice from JSON data

lookup_participant

Check if a company is on the Peppol network

get_invoice_status

Track delivery status of a sent invoice

validate_invoice

Validate invoice data against EN 16931

FAQ

Common questions

What does "AI-agent first" mean?

It means every interaction with peppol.sh can be done via API, without requiring human intervention. AI coding agents can create accounts, send invoices, and manage the full lifecycle programmatically.

Can my AI agent really create an account automatically?

Yes. POST to /v1/signup with an email address and you get a sandbox API key back immediately. No email verification, no CAPTCHA, no dashboard signup flow.

What is SKILL.md?

SKILL.md is an open standard for AI agent capabilities. Our SKILL.md file exposes peppol.sh tools that AI agents can discover and call directly. This includes sending invoices, looking up participants, and checking delivery status.

Does my AI agent need to understand Peppol/UBL?

No. Send simple JSON with invoice data (customer, line items, amounts). We handle the conversion to Peppol BIS 3.0 UBL, validation against EN 16931, and delivery via the Peppol network.

Is sandbox testing free?

Yes, forever. Sandbox invoices are delivered via email (not Peppol network) and are completely free. You only pay when you go live and send via the real Peppol network.

How does the AI agent go live?

After testing in sandbox, submit KYC details via /v1/account/kyc. Once verified (usually within 24 hours), create a live API key and start sending via the Peppol network.