Integrios is an open-source, self-hostable backend integration platform. It gives engineering teams a durable boundary for receiving events, tenant-aware routing and transformation, and reliable delivery to downstream systems, with retries, dead-lettering, replay, and end-to-end observability. Run it yourself and adapt it to whatever you need to integrate.
Note
Early preview, built incrementally. The backend foundation works end to end and is backend-first (an admin UI is planned). Evaluate it as a self-hostable foundation, not a turnkey production deployment. MIT licensed.
Not yet, but planned:
- Authenticated delivery. Today the worker delivers only to open, no-auth endpoints; authentication to real systems and more built-in integrations are coming.
- No admin UI. All configuration is through the Admin API.
- No RBAC or rate limiting yet. Tenant isolation is enforced at the data layer.
- Durable event intake behind a transactional-outbox acceptance boundary, so no accepted event is lost
- Tenant-scoped API-key authentication and isolation
- Topic/subscription routing with optional JSONata payload transforms
- Reliable async delivery: bounded retries, dead-lettering, and replay
- Per-event status and delivery-attempt history
- Pluggable, vendor-neutral observability (OpenTelemetry metrics, logs, traces); bring your own backend
- Clean control-plane / data-plane / worker separation; scales horizontally
- Self-hostable and Docker-first
Integrios splits platform intent from runtime execution. The control plane (Integrios.Admin) owns tenants, integrations, connections, topics, and subscriptions. The data plane takes over at runtime: Integrios.Ingress validates, authenticates, and durably accepts events behind a transactional outbox; Integrios.Worker fans out to subscriptions, applies transforms, delivers to destination connections, and handles retries, dead-lettering, and replay.
For the full design (processing flow, durability guarantees, and platform concepts), see docs/architecture.md.
Prerequisite: Docker.
cp .env.example .env
make upThis starts the services, Postgres, migrations, and a test sink (Admin API on http://localhost:5150, Ingress on http://localhost:5231). Then follow the setup guide to onboard a tenant and send your first event end to end.
- Setup & quickstart: run locally and deliver your first event
- Architecture: design, processing flow, and platform concepts
- Observability: metrics, traces, logs, and OTLP export
- CI/CD: the pipeline and published images
- Contributing
| Area | Technology |
|---|---|
| Language / Runtime | C# / ASP.NET Core (.NET 10) |
| Database | PostgreSQL |
| Event backbone | PostgreSQL transactional outbox + bus behind IEventBus (Kafka swappable when justified) |
| Observability | OpenTelemetry (OTLP-capable); Prometheus + Grafana for local dev; bring your own backend |
| Deployment | Docker / Compose; container images on GHCR |
- A central ingress hub for webhook-heavy ecosystems (payments, CRM, support, commerce, internal systems)
- Tenant-scoped fan-out of one event stream to multiple destinations with per-subscription logic
- Reliable buffering and recovery during downstream outages or rate limiting
- Auditable event and delivery history for compliance and incident response
Issues and pull requests are welcome. See CONTRIBUTING.md.