Skip to content

Why Arqen? ​

Backend code becomes harder to operate when routes, storage, background work, and authorization are scattered across implicit conventions. Arqen puts those pieces in explicit application modules and gives them consistent runtime support.

Common problems ​

“The API works, but nobody can discover what it does” ​

Define typed tools and manifests with names, inputs, outputs, scopes, effects, and idempotency behavior. A human client, script, or agent can inspect the same description before calling an operation.

See Typed tools, Agent discovery, and Manifest contract.

“Background work is mixed into request handlers” ​

Use Thingd-backed queues and workers for email, indexing, synchronization, and other work that needs retries, leases, idempotency, or dead-letter handling.

See Durable jobs.

“Changing storage changes application code” ​

Use the ThingdBackend contract so domain services can work with memory, native Thingd, or an HTTP Thingd service. Start locally with memory mode and move to durable storage after the application shape is clear.

See Thingd integration and Configuration.

“Production behavior is hidden in startup scripts” ​

Use layered configuration, strict production validation, health and readiness checks, structured logs, and explicit worker settings. The application still owns deployment policy, secrets, backups, and tenant isolation, but the checks and integration points are visible in the codebase.

See Deployment and the Production runbook.

When Arqen fits ​

Arqen fits a backend that needs a clear path from a small local service to a durable deployment. It can sit alongside a model runtime, a frontend, a workflow system, or another service without requiring any one of them.

It is not a model provider or a hosted database. Thingd provides the data engine; the application provides domain behavior; Arqen connects the HTTP, module, job, health, and storage pieces.

Current limits ​

Review Feature status before relying on a capability in production. Cloud storage, workload-specific recovery, key lifecycle, tenant policy, and backup operations require application and deployment decisions.

Rust-first backend toolkit · explicit integrations for HTTP, jobs, and Thingd