At a glance
Who this is for
Teams with a working API and no user management in front of it
Data providers who want to sell access, not build a billing system
Companies whose internal API now has to face paying customers
AI and document services that bill tokens, pages or compute time
What you would otherwise build yourself
Your API works. What is missing is everything around it. You have endpoints. You do not have accounts, roles, plans, quotas, usage metering, invoices with VAT, payment collection or a pricing page, and your customers expect all of them before they pay. Every piece also has to be maintained for as long as the product lives.
What you get
Access control on every route
Give the relay your upstream base URL and an OpenAPI document (3.0, 3.1 or Swagger 2.0). A protected document is fetched with the upstream credential. Each path then gets four named permissions: read, create, update and delete. You grant them to a user, an organization unit or a role, or let a subscription do it: buying a plan grants its roles, and ending or suspending it takes them back. A customer can read a path without being allowed to write to it. Access is decided before the request reaches you.
Plans, metering and billing
Attach the relay to a product and define plans. The gateway counts calls, data volume, credits or tokens and enforces plan limits: past a limit it refuses calls before your API sees them, or you bill overage up to a spend cap. Invoices with VAT, e-invoices and Stripe payment are included.
Your own route prefix
Routes are published under /eps/{your-prefix}/…: one endpoint per OpenAPI path and access level, not a catch-all proxy. The prefix is one path segment of up to 100 characters (letters a to z, digits, dot, underscore, tilde or hyphen), unique across the whole deployment whatever its case, so no other tenant can claim the same public path. A path your spec does not list, or that cannot be routed safely, answers 404. That is why permissions, limits and usage metering work per route, operation or relay.
Upstream credentials stay with you
UniCore injects the upstream credential on the way out: API key, bearer token, basic auth or a custom header. You rotate it in place without touching the relay or its grants. It is stored encrypted, never returned by the API and never sent to a caller.
A public Swagger page per relay
Switch on a relay's public Swagger page and it lists only the operations you serve, with UniCore's bearer sign-in wired in and metered operations marked. Upstream addresses and vendor extensions are removed.
Routes your spec doesn't advertise
Add manual endpoints for the parts of your API that your OpenAPI document does not cover. Each can forward to a different upstream path and use its own upstream credential. Callers are checked as on any other route.
Bill what your API reports
Tell the relay where to find tokens, pages or compute time: a response header, a JSON field or a stream event. UniCore books them per call against the customer's balance, also for streamed answers.
What ships with UniCore and what you write
Everything on the left is part of UniCore and maintained by us. Everything on the right is yours to write.
Ships with UniCore
Accounts, sessions and session renewal, with email and password sign-in
Sign-in with Google, Microsoft or your own OIDC provider (Platform package)
Roles, organization units and named permissions, down to the single record, plus access control per route
Tenants and tenant isolation
Products and plans with base price, trial and term, plus credit packs for top-ups
Usage metering per customer and quotas on calls, data volume, credits and tokens, plus rate limits and overage billing with a spend cap
IP allow-lists, audit of changes and reads (Platform package), and a request id on every relayed call
Upstream credential injection and rotation, and a test connection before you save a relay
An optional public Swagger page per relay
Streamed answers and opt-in WebSockets, with optional retries and a circuit breaker
An operator dashboard in German and English
Invoices with VAT from the nightly billing run, e-invoices (ZUGFeRD, Factur-X, XRechnung), credit notes, card payment through your Stripe account or a bank transfer you record, and dunning for failed payments
Storefront components in React for pricing, checkout, usage and invoices, fed by the live catalog through a typed API client
API keys your customers create themselves
Usage booked from what your API reports
You write
Your API
Your website, with the React storefront components for pricing, checkout and invoices
Your prices, your seller and tax details, and legal texts your counsel has approved
Frequently asked questions
Your question isn't listed? Tell us briefly what it's about.
Your API stays as it is. The relay sits in front of it and forwards to your upstream base URL. Access, plans, limits and call and byte counts work with the API as it is today. To bill tokens, pages or compute time, the relay reads them from a JSON field or stream event your API already returns, or from one response header you add.
There are three modes for that. You work with the spec only, with the spec plus manual endpoints (the default), or with manual endpoints only, with no spec at all. That gets you there even when the spec has gaps.
The Relay package. It includes relays, plans, metering, invoices, payment, jobs and webhooks. Customers sign in with a password or an API key. Build adds your own data models and pipelines. Platform adds sign-in through Google, Microsoft or your own OIDC provider, messaging channels, time series and auditing.
Quotas apply to users who reach the route through a plan they subscribed to. A user who calls it through a permission granted directly or a default role stays free of quota and overage. For those callers, the rate limits you set on the relay or endpoint apply. Roles and permissions decide access, quotas govern usage.
See it applied to your own use case
Thirty minutes, screen-shared: a relay, a plan, a metered call and an invoice, then your API or product idea. No slide deck.