Back to Projects

Radon

2026

Provider-Agnostic Backend SDK Suite

R
TypeScriptSDKNode.jsAuthPaymentsOpen Source

The Challenge

Every product ends up gluing together five separate vendor SDKs for auth, payments, storage, email, and AI — each with its own auth scheme, quirks, and webhook shape. Switching providers later (Stripe to Paystack, S3 to R2) usually means a rewrite, not a config change. Most existing solutions solve this by becoming a hosted service with per-user pricing and full data lock-in.

The Solution

Built Radon as five SDKs sharing one interface shape: auth, payments, storage, email, and AI, each normalizing 10-50+ providers behind a single typed call (auth(), charge(), upload(), send(), chat()). Provider-specific quirks are absorbed inside each adapter and never leak into calling code. Everything runs against the developer's own infrastructure and provider keys — no hosted service, no stored secrets, no per-MAU pricing. Escape hatches (native()/rawClient()) cover the 20% of cases the unified API doesn't.

Impact

  • •Shipped 126+ provider integrations across 5 SDKs with zero runtime dependencies
  • •Free tier is genuinely production-grade (not a crippled trial) — full MIT-licensed core across all 5 SDKs
  • •Pro tier priced as a single $7.99 lifetime license unlocking every SDK, unlimited projects — no subscription
  • •Built adapters lazy-load, so an app using only OpenAI + Anthropic never bundles the other providers
  • •Designed the suite to expand (payments, storage, AI added after auth validated the model)

Key Features

One typed interface per domain (auth, payments, storage, email, AI) across 10-50 providers each

Provider swap by config change, never a code rewrite

Self-hosted — no account, no hosted service, runs against your own DB and provider keys

Escape hatches to the native provider client for edge cases

Zero required runtime dependencies, ESM + CJS, strict TypeScript

Single lifetime license unlocks Pro across the entire suite

Tech Stack

TypeScript, Node.js ≥18, ESM + CJS dual build
Framework-agnostic (Next.js, Express, Fastify, Hono, NestJS, and more)
Database-agnostic auth adapters (Postgres, MySQL, SQLite, MongoDB, Firebase, and more)
Normalized webhook handling across payment/email providers

Lessons Learned

  • →Generalizing auth logic already built twice (in two prior projects) into a real SDK surfaced most of the hard edge cases early
  • →A lifetime one-off price removes the friction of a subscription decision for indie developers
  • →Keeping the free tier fully complete (not a trial) is a stronger trust signal than a generous trial period
  • →Scoping v1 to JS/TS only, deferring other languages, kept the "zero hosting, zero liability" pitch intact

Interested in working together?

Let's build something amazing.

Get in Touch