Introduction
Kortecx is the distributed runtime for AI agents — the execution kernel beneath them. Not a model, not a database, not a chat app.
Kortecx is the distributed runtime for AI agents. Not a model, not a database, not a chat app — the execution kernel beneath them. Describe the work; the runtime takes over.
v0.1.0 ships today, open-source under Apache-2.0, built in the open. It installs as one FFI-free binary — kx — and runs end-to-end on a laptop. The same runtime scales out to a coordinator/worker layer without a rewrite.
What you get
- Exactly-once delivery with transparent recovery. Crash mid-commit and replay: committed work is re-read, not re-run.
- An append-only journal as the single source of truth, with deterministic replay.
- Content-addressed motes — each step is one immutable, content-addressed unit of work.
- A policy-gated CapabilityBroker — the single door to world effects, with a refusal gate that rejects unsafe work up front.
- Local model execution — optional llama.cpp inference behind a trait seam, GGUF by path.
- An embedded gRPC gateway + event WebSocket via
kx serve.
Where to next
- Quickstart — prove exactly-once in 60 seconds.
- Install — curl, cargo, or Docker.
- Core concepts — mote, journal, projection, commit protocol, refusal gate, CapabilityBroker.
kxCLI reference — every command and its flags.- Recipes — five reusable workflow shapes.
- Architecture — the 39-crate layered DAG and the optional distributed layer.
Open-source, install-now
The runtime is free to self-host. The managed cloud (hosted tier) is on the waitlist; enterprise and services are contact-sales.