Platform & engine

ВекторХаб and Вектор

ВекторХаб is repository hosting and team collaboration around code. Вектор is distributed version control. They are compatible only together: interoperability is designed for the ВекторХаб + Вектор pair, not separately.

Installers

Download Вектор CLI

Pick your platform. When your operator sets download URLs in the environment, each card becomes a direct link.

Linuxvector-x.y.z-linux-amd64.tar.gzDownload

Windowsvector-setup-x.y.z-x64.exeDownload macOSvector-x.y.z-macos-universal.dmgDownload

What you use

Product surface: in the browser and on the command line with the client tooling you already use.

ВекторХаб

This product

A source-code workspace for hosting repositories, reviewing changes, and coordinating teams — with the ergonomics of a modern forge.

Repositories & workflow
Repositories on ВекторХаб with access over HTTPS and SSH, issues, pull requests, reviews, labels, milestones, branch protection, webhooks, commit statuses, and an audit trail.
Organizations & access
Teams, collaborators, roles, stars, watches, and notifications so the right people stay in the loop.
Security built in
Two-factor authentication, personal access tokens, SSH keys, and session-aware APIs behind the same account model.

Вектор

Product line

A standalone product line for distributed versioning and code collaboration on a home-grown stack—focused on the engine and data model, not on borrowing someone else's hosting chrome.

Recognized concepts
Commits, objects, packed history, merges, and collaboration primitives in familiar semantics—with room to experiment under the hood.
Separate codebase
Вектор evolves on its own timeline. ВекторХаб can interoperate over the wire with mainstream tooling today while Вектор advances its own stack.
Why keep it separate
Shipping a forge and redesigning a VCS core are different risk profiles. Вектор can iterate on algorithms and wire protocols while ВекторХаб focuses on product workflows customers expect today.

How ВекторХаб is built

Concrete stack, storage layout, and request paths — not marketing fluff.

Browser & BFF

The UI is Next.js (App Router). Sensitive calls go through same-origin `/api/*` routes so HttpOnly cookies and CSRF tokens never need to live in public JavaScript against the raw API origin.

API & metadata

FastAPI exposes versioned REST under `/api/v1` with OpenAPI at `/docs`. SQLAlchemy maps users, repositories, issues, pull requests, permissions, webhooks, audit events, and more into PostgreSQL. Schema changes ship through Alembic migrations.

Vector VCS on disk

Each repository is a Vector object store on the filesystem (owner/name.vector). Server-side browsing, diffs, and merges use the native Vector format and remote protocol.

Typical signed-in request

1 · Page load
React components call `fetch` to `/api/...` on the same origin. The BFF forwards cookies, injects anti-CSRF headers where required, and proxies to the FastAPI base URL.
2 · Authorization
FastAPI dependencies decode JWT access tokens (with refresh rotation), resolve repository roles, and reject under-protected routes before any VCS work runs.
3 · Domain work
Handlers read or write PostgreSQL, enqueue outbound email or webhook deliveries, and read or write Vector object stores when the operation touches commit data.

How the pieces relate

ВекторХаб answers where teams host, review, and ship. Вектор asks how a version-control core could be implemented and extended. You can adopt the hub without running Вектор, and study Вектор without using this site.