finds.dev← search

// the find

panoratech/Panora

★ 1,032 · TypeScript · AGPL-3.0 · updated Oct 2025

Open-Sourcing the integration engine our platform runs on

Panora was a unified API platform — think Merge.dev or Nango — that let you connect your app to CRMs, file storage, ticketing systems, and e-commerce platforms through a single normalized API. It's a NestJS monorepo with a React dashboard, magic-link OAuth flow, and an embedded catalog widget. The OSS product has been sunset; the README is now one sentence pointing to the commercial product.

- The architecture was well-thought-out: NestJS backend with Prisma, a separate magic-link frontend for end-user OAuth consent, an embeddable React catalog widget, and a Speakeasy-generated SDK — each concern in its own app rather than one bloated monolith.

- Self-hosted path was real: Docker Compose setup with health checks, multiple Dockerfiles per app (dev/prod/slim variants), and a proper .env.example. Not the usual 'self-hostable in theory' situation.

- Field mapping was a first-class feature — the UI has a stepper for defining and mapping custom fields across connectors, which is the thing most unified-API competitors either skip or hide behind enterprise plans.

- Changesets + automated release workflow meant SDK versioning was tied to backend releases, so clients weren't chasing undocumented API drift.

- It's dead. The README is a tombstone. No amount of good architecture matters when the maintainers have moved on and the OSS layer is explicitly abandoned — you'd be forking a ghost.

- No meaningful documentation in the repo itself — no architecture doc, no connector development guide, nothing that would let someone pick this up and extend it. If the team stops answering questions, you're reading source code cold.

- The monorepo has significant duplication: the magic-link app and the embedded-catalog app share nearly identical component trees (Modal.tsx, form components, OAuth hooks) with no shared package between them. A sign that contributors were moving fast and nobody cleaned up.

- Connector coverage was uneven — CRM and file storage got real attention, but the ecommerce and accounting verticals looked thin based on the asset directories, which matters if your use case isn't Hubspot or Google Drive.

View on GitHub → Homepage ↗

// want more like this?

We dig through GitHub every week and send a few repos picked for what you actually care about — each with an honest take like this one.

Get finds in your inbox → Search again →