Typed tools
Make application capabilities discoverable with structured inputs, outputs, permissions, and audit metadata.
Typed tools, durable jobs, discoverable APIs, and thingd integration for services that people, programs, and agents can operate.
Early-stage project
Arqen is Rust-first and actively maturing. The core framework and CLI are implemented in one package, but production adoption still requires validating durability, public thingd compatibility, security, and operational behavior for the application being built. Check the feature status before adopting a capability.
The current implementation uses Rust, Tokio, Tower, tracing, and native thingd adapters, with Axum as an internal HTTP transport. The application positioning stays language-agnostic: future Node.js support can use the public HTTP API, SDKs, templates, and shared manifests.
cargo run -p arqen --features cli --bin arqen -- new hello-api
cd hello-api
cargo runOr run the workspace server directly:
cargo run -p arqen --features cli --bin arqen -- dev --storage memory
curl http://127.0.0.1:8888/healthclient or agent → Arqen API → typed tools / policies / jobs / logs
↓
memory · native durable thingd · HTTP sidecar · future cloudApplication code should depend on domain interfaces and the public thingd contract, not private cloud modules. See architecture and thingd integration.
thingd supplies the storage, events, search, links, and queue boundary that Arqen adapts. Arqen keeps that integration optional and public-contract based; cloud hosting is a future path rather than a current promise.
Learn more about the thingd ecosystem at thingd.cloud, the home for thingd’s hosted data engine and related services.
Arqen is not another model runtime or hosted backend. It is the contract layer between an application and the software that operates it:
Read Why Arqen? for the tradeoffs and the honest maturity boundary.
Read the roadmap and the Phase 17 specification on GitHub for implementation evidence and open work.
Read the performance guide for benchmarks, profiling, and optimization patterns.