Roadmap

Where ForgeOS stands, and what it refuses to claim.

A roadmap is a set of claims about the future, so it follows the same rule as everything else here: proven things are marked proven, intentions are marked intentions, and what does not exist is said not to exist.

Cited release v1.20.1

Now, next, and later

Everything below is stated as of v1.20.1 and traces to the repository's own roadmap and status record.

shipped released at the cited version and verified — every claim is checkable in the repository

in progress underway now; described as intent, never claimed as done

deferred deliberately not being built yet, each with its blocker or reason named

Now

Command layer and project intake

shipped

A command layer that answers the question asked at the start of a session: where is this project, and what is the next safe thing to do
Project ingestion — a project's own documents become compact, cited maps, read before implementation
A read-only intake report and readiness forecast that classify a project and name the next prompt family
Closure integrity and a self-measuring session brief (v1.20.0), then a self-test that returns the same verdict on any Windows console and a project-owned declaration of generated build outputs (v1.20.1)

Real-project proof

in progress

Driving projects end to end through the full work loop, beyond the repository's own slices
Ingestion slice 2 — surfacing ingestion state through the command layer
Ingestion slice 3 — gating validation for a documented project with no maps, only once adopters have proven it

Later

Deliberately deferred

deferred

Package-manager channels — each deferred with its blocker named
A documented compatibility matrix of which adopted versions can upgrade to which — still outstanding
A read-only dashboard, only after the machine-readable status contract proves stable

Not yet proven

Design intent stated as design intent. Every line below is listed by the repository itself as unproven until real use proves it.

macOS: the POSIX half is expected to run there and has never been tested on it. CI runs Windows and Ubuntu.
The full work loop on a second project. It has run end to end on this repository's own slices, and three cold sessions on one adopted project each closed a slice; no second adopted project has done the same.
Three of the roles have never been dispatched on real work.
Whole-session efficiency: only the session launcher is measured. Nothing yet states what a whole managed session costs against an unmanaged one.
Closure on a real adopted project: the v1.20.0 closure behaviour is pinned by fixtures; the operational field trial is still pending, and the acceptance behind it was not independently reviewed.
Scale: this has been proven on real projects, not on many projects at once.

Deliberately not built

Absent on purpose, not by omission — each with its reason on the record.

No package-manager install channels yet — every channel is deferred with its blocker named, and there is no curl-into-a-shell by design.
No dashboard — a read-only one comes only after the machine-readable status contract proves stable.
No hosted service, paid tier, or team governance product while the open core is still proving itself in public.
No telemetry or analytics of any kind — nothing is collected, so there is nothing to opt out of.

What is not yet proven is listed as not proven. The full roadmap — including what is deliberately not being built — lives in the repository. Read the full roadmap in the repository.