Roadmap

Read the maturity column before deploying anything

Nothing in this ecosystem is production-ready. Kallisto’s own README is blunt about it: “Do not run this where it matters.” Treat every project here as research code until its documentation says otherwise.

Current state

Project	Language	Maturity	What works today
Kuberina	Rust solver, Go frontend planned	Alpha	Rust solver: FFD warm-start, GA optimisation, CSP forward checking. Benchmark and validation pipeline via Make targets.
Helvilette	Go	Alpha	Othela control plane, polling agent, Git clone and ansible-playbook execution, structured JSON reporting, E2E suite.
Kallisto	Rust 2024	Prototype	KV-v2 read/write path, cuckoo cache, CLOCK eviction.

Known gaps

These are the things most likely to bite you, stated plainly.

Kuberina — the kuberina Go binary is currently a stub. The CLI described as kuberina plan in the project README is the planned frontend (kuberina-forge), not shipped code. What runs today is the Rust solver, driven through the Makefile. See running the solver.

Helvilette — the project is pre-release. Contributor guide, mailing lists and communication channels are still placeholders in the upstream repository.

Kallisto — not built yet: authentication on the data port, TLS, the encryption barrier, and the entire controlplane. The dataplane is the only part that exists.

Planned work

Kuberina

Phase 1 — Forge and hexagonal architecture (v0.3.0). Build the kuberina-forge Go CLI as a compiler-style frontend and backend: ingest raw manifests, Helm charts, Kustomize output and cloud API state into a standardised Kuberina IR, then render the optimised blueprint back into executable manifests. Refactor the Rust solver onto ports and adapters so the FFD/GA/CSP domain core is fully decoupled from I/O.

Phase 2 — Distributed solving and agnostic targets (v0.4.0). Replace the YAML exchange between forge and solver with streaming gRPC over Protobuf. Because the solver only ever sees Kuberina IR, extend it beyond Kubernetes to Proxmox, Nomad, Slurm and IoT edge provisioning. Run lightweight forge agents in expensive cloud environments while the heavy genetic-algorithm computation happens on cheap on-premise hardware.

Kallisto

Pluggable storage backends beyond the RocksDB reference implementation — SQLite and other key/value systems. Docker Engine secret storage support.

Version numbering carries a warning worth repeating: releases 1.0.0 through 2.0.0 are not production releases, and 1.0.0 begins the rewrite into Rust, so breaking changes are expected. 2.0.0-lts is intended to be the first production-ready tag.

New projects

Kalena, Kaeliir and Ginnungagap are planned members of the ecosystem. They will get their own documentation sections here as they become public — the site structure already accommodates them, and adding one is a single entry in the sidebar configuration.

Licensing trajectory

Kuberina publishes an unusually explicit three-phase licensing plan, worth understanding before you build on it:

1. AGPLv3 from 0.1.0 up to (not including) 12.0.0.
2. SSPLv1 if and when the author’s corporate entity is formally incorporated and the IP transfers to it. Already-released AGPLv3 versions stay AGPLv3 permanently.
3. Apache 2.0 on either a CNCF donation initiative or reaching 12.0.0 — and in any case automatically, 48 months after each release’s own date.

Kallisto is AGPLv3, with commercial licensing negotiable. Helvilette is Apache 2.0. Full detail lives in Kuberina’s licensing reference.