fix: a token-gated or LAN-bound dashboard works end to end (WSL mirrored)

- /ws-pty carries the dashboard token like /ws (the terminal never attached
  with DASHBOARD_TOKEN set), and so does everything that cannot send the
  header: images, downloads, exports, drops (withToken/tokenHeader). A
  refused upgrade answers 401/403 and is logged and counted by reason.
- The browser asks for a missing or wrong token (401 EUNAUTHORIZED) and takes
  ?token= from the link agenthub start prints, dropping it from the URL.
- The CLI sends the token (DASHBOARD_TOKEN, else the one the running server
  publishes 0600 in <data dir>/runtime/dashboard-token); MCP and statusline
  read the same file.
- DASHBOARD_HOST takes a list, and 127.0.0.1 is always listened on too, so a
  LAN-only bind keeps hooks and the CLI working.
- agenthub start --host/--allowed-hosts/--token are saved in
  <data dir>/network.json for every later start; a wide bind gets a token;
  every reachable URL is printed.
- tmux < 3.7 is an amber, once-per-version notice that says only the pane
  scrollbar is missing, with a ~/.local build instead of sudo make install.
- INSTALL: PATH belongs in ~/.profile for sudo -iu/ssh; WSL mirrored notes.

docs/CLI.md

@@ -99,7 +99,7 @@ The CLI talks to the local dashboard server — **API-backed commands require it
 | Command | Description |
 | ------- | ----------- |
 | `agenthub status` | At-a-glance up/down indicator (`●` running / `○` not running); exits `1` when down |
-| `agenthub start [--port N]` | Start the production server **in the background** (detached; survives closing the terminal), wait up to 30 s for `/api/health`, print the URL + PID and the `agenthub stop` hint. Logs append to `data/agenthub-server.log`. No-ops with a pointer when a server is already up. Requires a built client (`npm run build` once). Always ends a Quit (`agenthub stop`); in a plugin install it starts the server the way a session does — by running `scripts/plugin-bootstrap.js` |
+| `agenthub start [--port N] [--host A[,B]\|local] [--allowed-hosts H,..\|none] [--token T\|auto\|off]` | Start the production server **in the background** (detached; survives closing the terminal), wait up to 30 s for `/api/health`, print the URL + PID and the `agenthub stop` hint. Logs append to `data/agenthub-server.log`. No-ops with a pointer when a server is already up. `--host`, `--allowed-hosts` and `--token` are saved in `<data dir>/network.json` (0600) and applied by every later start (a set environment variable wins); a non-loopback `--host` without a token gets one generated. The links to open — every address of a `0.0.0.0` bind, each with `?token=` — are printed. Requires a built client (`npm run build` once). Always ends a Quit (`agenthub stop`); in a plugin install it starts the server the way a session does — by running `scripts/plugin-bootstrap.js` |
 | `agenthub stop` | **Quit AgentHub** (issue #65): closes every workspace terminal (`agenthub-lane-*` tmux sessions — running agents stop), stops the server — through `systemctl --user stop` / `launchctl bootout` when the autostart unit runs it — and keeps it stopped: Claude Code's `SessionStart` bootstrap starts no server while `<data dir>/runtime/stopped` exists. `agenthub start` (or any server starting, e.g. a boot/login with autostart on) clears it. With the server already down it only sets the flag |
 | `agenthub uninstall [--yes]` | **Uninstall AgentHub** (`scripts/uninstall.js`; Settings → System has the same button). Without `--yes` it only lists what is there and exits 1. With it, in order: closes every workspace terminal, removes the systemd/launchd unit, stops every dashboard server, the status-area icon and workspace stacks, takes AgentHub's own entries out of `~/.claude/settings.json` (hooks, statusline; backup `settings.json.agenthub-uninstall.bak`), Codex `hooks.json` and Antigravity `hooks.json`, deletes the AI Proxy `$CODEX_HOME/<name>.config.toml` files, runs `claude plugin uninstall` for `agenthub@*` and every plugin of the `agenthub` marketplace and removes that marketplace (never a shared one), then deletes `~/.local/bin/agenthub`, `~/.config/agenthub/`, `~/.claude/.agenthub.json`, `~/.claude/agenthub-{proxies,routers,git-credentials,accounts,profiles}`, `agenthub-lanes/.state` and `.locks`, the plugin's cached copies and the data directory (database included). **Kept:** the AI tools' own conversations (`~/.claude/projects`, `~/.codex/sessions`, …), workspace worktrees and branches, anything in your repositories, config backups. A data directory that is the home, `CLAUDE_HOME`, a checkout or not recognisably AgentHub's is refused. A failing step is logged and the rest still run; log: `$TMPDIR/agenthub-uninstall-<uid>.log`. Needs no running server. Run inside a dashboard terminal, it closes that terminal last. |
 | `agenthub tray` | Show AgentHub's **status-area icon** now, in the foreground (`scripts/tray.js`): Open AgentHub, Quit AgentHub (a submenu asks again, then runs `agenthub stop`), Start AgentHub after a quit (`agenthub start`), Hide this icon. A production server already keeps the icon up on its own unless Settings → System turned it off; this command does not change that setting. Exits at once when an icon is already running |