Feature

One page over every board

in this fork since v2.22.0

A board that can reflash four computers should not face the internet. One page inside the cluster drives them all, with every control a board has, and it holds no credential of its own.

1hostname exposed for the whole estate

11pages per board, the same ones a board serves

0credentials held by the page

hostname exposed for the whole estate: the fleet interface as shipped in BMC-UI 3.29.0, 2026-09-13. pages per board, the same ones a board serves: the fleet interface as shipped in BMC-UI 4.0.0, 2026-10-03. credentials held by the page: the fleet interface as shipped in BMC-UI 3.29.0, 2026-09-13.

See it in the demo →Two boards on one page, from a real board's capture.Do it on your board →Install the firmware each board in the fleet runs.The evidence →What is still not fixed, including where trust is assumed.Who may reach this board →The two ways in, and what a trusted proxy buys.

Both boards on one page, each answering for itself. A board that is down costs you its card and nothing else.

The argument

The page renders each board's own pages from the same data layer the board's page uses, so the two cannot drift apart. It holds no credential: a gateway in front of it authenticates the operator and presents a client certificate the board trusts, which is why no board needs to be reachable from anywhere else.

The full argument, with every measurement →