Skip to content

[NET-01] Evaluate optional NetBird private-network integration #19

Open

Description

@jdbencardinop

ID: NET-01. Lifecycle: requested only. Backlog: #1.

The maintainer uses NetBird. Evaluate optional NetBird-aware endpoint
advertisement/exposure without requiring Tailscale or adding another federation
layer.

What works already

Generic authenticated direct pairing can use a reachable private-network IP.
Running the fork on a NetBird-connected host and binding/advertising its private
address may be sufficient; clients can pair manually. Mesh routing does not
replace T3 authorization, and HTTPS-hosted clients still need a compatible HTTPS
backend.

T3's Tailscale integration reads local status/MagicDNS, advertises local
endpoints, manages Tailscale Serve HTTPS and probes the T3 well-known endpoint.
It does not enumerate or auto-connect to every Tailscale peer. The existing
AdvertisedEndpoint/AdvertisedEndpointProvider types are a reuse seam, not a
generic WireGuard connector API.

Important differences to preserve

  • Current desktop code labels all 100.64.0.0/10 interface addresses as Tailscale.
    NetBird can use that range, so product identity cannot be inferred from it.
  • NetBird's local status/API can support discovery of this host's capabilities;
    no daemon install/login/reconfigure/restart is authorized by this request.
  • netbird expose publishes to the public internet and is not equivalent to
    private Tailscale Serve. Public exposure must be a separate explicit scope.
  • Headscale is a Tailscale-compatible control server; verify DNS/certificate/
    Serve support instead of promising interchangeability.

Prefer the manual path first. If useful, add an optional capability-based
endpoint provider, off by default, preserving current Tailscale behavior,
environment-scoped ownership and ordinary operation without any VPN.
Peer inventory, T3 service discovery, TLS exposure and pairing are distinct.

Recorded request

No runtime implementation, phase advancement, VPN mutation or live reachability
test. This is a proposal for evaluation, not shipped NetBird support.

Recorded by GPT-6 Astra through GitHub Copilot CLI; no delegated inference.

Activity

  1. added
    fork-backlogMaintainer backlog for the multi-machine fork
    priority:P1Next workspace capabilities after foundation
    on Oct 2, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    fork-backlogMaintainer backlog for the multi-machine forkpriority:P1Next workspace capabilities after foundation

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions