Maturity
How far each part of Neuralis is today — what the maturity labels mean and how a score is formed.
Neuralis is in public beta, and not every part of it is equally far along. Each package and each cross-cutting topic — an install channel, a platform, sign-in, MCP, the marketplace — carries a maturity label and a score that say how much of what it promises and plans is built today. The same label appears at the top of the page that documents the subject.
What the labels mean
| Label | Score | What it means |
|---|---|---|
| stable | 85 and above | The contract is settled for the release line and changes only with notice; no subject carries this label yet. |
| beta | 65–84 | Complete for its documented contract and working end to end; the contract may still change with notice, and known limits are listed. |
| preview | 40–64 | Usable, with documented gaps; expect change. |
| experimental | below 40 | Present but incomplete or unverified; do not depend on it. |
| not available | — | Described, but nothing of it ships yet. |
How a score is formed
A score measures how far a subject is developed against what it promises and what is planned for it. Every subject is broken into fine-grained capabilities — a contract field, a tool, a screen, a connector, an operator command — and each capability belongs to one of three tiers:
- Promised — what the documentation, the package contract and the README say the subject does today. Counts in full.
- Planned — work planned before a stable release. Counts in full.
- Vision — a few long-term directions that are clearly this subject's. Count a quarter.
Each capability then counts as built (in full), partly built (half) or not built yet (nothing). The score is the built share of the whole, weighted by tier and rounded down to a multiple of five — so a subject with a long plan scores lower until the plan is built. Planned and vision work enters the count only by an explicit decision on each item.
A capability is partly built only when a promised part of it is missing, or when an open high-severity issue makes that capability unusable or unsafe. Open issues never lower the score by how many there are; they are fixed in the normal course of work.
A topic — an install channel, a platform, sign-in, MCP, the marketplace — is a view over the capabilities of every package that serves it. A capability counts once in each topic it belongs to and is never copied.
Caps
The score sets the band; four caps can hold the label lower whatever the score, which is why a subject can score 85 or more and still read beta:
- A subject with an open high-severity security issue on one of its capabilities is at most beta.
- A platform or channel that has never run on a live installation is at most experimental.
- A subject that cannot be obtained publicly — an install channel, the marketplace — is at most preview.
- Until the project runs continuous integration, every subject is at most beta.
Reading "planned" and the limits
"Planned" means the capability does not exist today. It is counted so the score shows the distance to the promise; it carries no date and is not a commitment to one.
The limits listed for each subject are lasting boundaries of how it works today — not a list of open bugs.
The team re-verifies these numbers for each release.
Every subject
Maturity: how far each part of Neuralis is today, as of 2026-10-05.
| Subject | Kind | Label | Score | Main limit |
|---|---|---|---|---|
| Package system | package | stable | 85 % | Source packages are markdown only; code from a synced source runs only when an owner flags that source as a package source, and then in the WebAssembly sandbox. |
| Agent core | package | beta | 80 % | Confined shells need a Linux host kernel with Landlock support; without it, lower-trust shells are refused rather than run unconfined. |
| Brain core | package | stable | 90 % | Git remote operations use HTTPS only; SSH remotes are not supported. |
| Admin | package | stable | 85 % | Admin adds no agent tools or slash commands; agents reach administration through its skills and the routes they wrap. |
| Machine core | package | stable | 90 % | Machines need access to a Docker engine; without one the platform runs with no machines. |
| Host (neuralis) | package | stable | 90 % | Sign-in is email and password; single sign-on (OIDC, SAML) is not available. |
| Install channels | topic | preview | 50 % | No install channel is publicly available; installs are built from the source repository. |
| Platform: Linux | topic | experimental | 85 % | No run on a standalone Linux host is recorded; the live-tested setup is Docker on WSL2. |
| Platform: Windows with WSL2 | topic | stable | 90 % | Use an in-distro Docker engine; with Docker Desktop the host broker needs its loopback TCP fallback. |
| Platform: macOS | topic | experimental | 35 % | Neuralis has not been run on macOS so far. |
| Platform: Windows (native) | topic | experimental | 35 % | Neuralis has not been run on native Windows so far. |
| Enterprise security and governance | topic | stable | 85 % | Give the host access plane only to a single trusted operator's machine. |
| Sign-in | topic | stable | 85 % | Single sign-on (OIDC, SAML) is not available; sign-in is email and password. |
| MCP | topic | stable | 85 % | Keep the MCP port private unless remote MCP clients are intended. |
| Workflows and scheduling | topic | beta | 75 % | Workflows run in a single process; a second replica would run every schedule twice. |
| Messaging channels | topic | preview | 55 % | Telegram and WhatsApp are the only messaging platforms. |
| Ecosystem import | topic | stable | 85 % | Imported content is markdown only: hooks and MCP server definitions in a foreign layout are not loaded. |
| Marketplace | topic | preview | 70 % | The marketplace is not publicly hosted; it runs as a local deployment only. |