Merge pull request 'fix(darwin): disable hot-corners' (#109) from fix/disable-hot-corners into main

Reviewed-on: #109

docs/shell.md

@@ -186,6 +186,14 @@ fleet's stock "wheel, with a password" policy. What it does **not** implement:
 host aliases, LDAP/SSSD sudoers, `sudoreplay`, and most `Defaults` settings.
 Needing any of those means reverting to `security.sudo`.
 
+One exception to the password: the EDaaS box sets
+`security.sudo-rs.wheelNeedsPassword = false`. NixOS-WSL ships that default for
+`security.sudo` — WSL has no console login, so the trust boundary is the Windows
+session and the Linux account password is never one the user chose — and the
+option does not carry across to the `security.sudo-rs` module, which defaults to
+requiring one. Without the explicit setting, `sudo` on that host prompts for a
+password nobody knows.
+
 If a host ever refuses to escalate, get a root shell that does not go through
 sudo (`wsl -u root -d NixOS` on the work box; the console or a serial/HDMI login
 elsewhere) and roll back with `nixos-rebuild switch --rollback`, or pick the
@@ -350,10 +358,10 @@ Claude to route new memories there.
 ## Per-host differences
 
 |                             | Personal Linux (sway) | macOS                 | Work WSL (EDaaS)             |
-| --------------------------- | --------------------- | --------------------- | --------------------------- |
+| --------------------------- | --------------------- | --------------------- | ---------------------------- |
 | Auto-tmux                   | yes (foot/TTY)        | yes (iTerm2)          | yes (WSL shell)              |
 | `kubectl` → kubecolor       | no (no kubectl)       | no                    | yes (work module)            |
-| `sudo` implementation       | sudo-rs               | Apple sudo + Touch ID | sudo-rs                     |
+| `sudo` implementation       | sudo-rs (password)    | Apple sudo + Touch ID | sudo-rs (passwordless wheel) |
 | git email                   | `iam@emmathe.dev`     | `iam@emmathe.dev`     | `…@citrix.com` (work)        |
 | ssh config managed          | yes                   | yes                   | no (keeps corporate config)  |
 | ssh-agent                   | yes                   | launchd               | yes (work module)            |

home/claude/memory/MEMORY.md

@@ -15,3 +15,5 @@
 - [Dev clusters disposable](dev_clusters_disposable.md) — Lyra's dev clusters are recreatable; mutate/break freely, no confirmation needed
 - [Nix shell tooling](nix_shell_tooling.md) — any nixpkgs tool runs ad hoc via `nix run`/`nix shell nixpkgs#<pkg>`; a missing command is never a dead end
 - [WSP local build and test](wsp_local_build_and_test.md) — core-services-cloud on this box: dotnet via nix, artifactory creds from `~/.artifactoryenv` sourced per command, how to tell auth failure from a code failure
+- [WSP-32957 PIM migration](wsp_32957_pim_migration.md) — AKS RBAC to PIM + AutoPerm decommission; prod is in the Technical Preview subscription, three tenants; resume via `~/code/WSP-32957-CONTINUATION.md`
+- [Entra group member reads](entra_group_member_reads.md) — `az ad group member list` hides service principal members; use the servicePrincipal cast or transitiveMemberOf, and trust the Terraform plan over it

home/claude/memory/entra_group_member_reads.md

@@ -0,0 +1,28 @@
+---
+name: entra-group-member-reads
+description: az ad group member list and Graph /members silently omit service principal members — use the servicePrincipal cast or transitiveMemberOf when a group is expected to hold an SPN.
+metadata:
+  node_type: memory
+  type: reference
+---
+
+`az ad group member list --group <id>` and `GET /groups/{id}/members` both return an **empty collection**, with no error, for a group whose only members are service principals — at least when called with Lyra's user account. The read looks authoritative and is not.
+
+**Reliable reads instead:**
+
+```sh
+# members, cast to the type that is being hidden
+az rest --method get --url "https://graph.microsoft.com/v1.0/groups/<gid>/members/microsoft.graph.servicePrincipal?\$select=id,displayName"
+
+# count, which does not filter
+az rest --method get --url "https://graph.microsoft.com/v1.0/groups/<gid>/members/\$count" --headers ConsistencyLevel=eventual
+
+# from the principal's side
+az rest --method get --url "https://graph.microsoft.com/v1.0/servicePrincipals/<spid>/transitiveMemberOf?\$select=id,displayName"
+az rest --method post --url "https://graph.microsoft.com/v1.0/servicePrincipals/<spid>/checkMemberGroups" \
+  --body '{"groupIds":["<gid>"]}' --headers "Content-Type=application/json"
+```
+
+**How to apply:** any group that deployment or automation identities belong to — `*-cluster-admins`, `*-keyvault`, anything created by `rg-prereqs` — must be checked with one of the above before concluding it is empty. Cross-check against Terraform: a plan reporting "No changes" against a `members` attribute is strong evidence the membership is present, and outranks the `az` read. On 2026-08-28 the plain read produced a Bug (WSP-33432, cancelled) claiming five `*-cluster-admins` groups across three tenants had been emptied; all five held their deployment principals the whole time.
+
+Related: [[wsp-32957-pim-migration]].

home/claude/memory/jira_wsp_fields.md

@@ -9,7 +9,9 @@ Field map for the **WSP (Workspace Platform)** Jira project, to create tickets w
 
 **Issue-type IDs:** Epic `10000`, Story `10004`, Task `10008`, Bug `10123`, Sub-task `10009`.
 
-**Fast path — use Task, not Bug.** A `Task` requires only `summary` (project/issuetype auto, reporter defaults to caller). A `Bug` requires six extra fields (below), so only pick Bug when it must be a Bug. The sibling infra/remediation tickets in WSP are Tasks.
+**Fast path — use Task, not Bug.** A `Task` requires `summary` plus **Task Type** `customfield_15622` (added since this note was first written; the create validator enforces it even though `createmeta` omits it, same trap as Bug's `versions`). Options value=id: Dev Task=34065, CQE Task=34066, Investigation=34067, Security=34068, Maintenance=34069, Release=34070 — use `Maintenance` for refactors and tidy-ups, `Dev Task` for feature work. A `Bug` requires six extra fields (below), so only pick Bug when it must be a Bug. The sibling infra/remediation tickets in WSP are Tasks.
+
+Example Task `additional_fields`: `{"customfield_15622":{"id":"34069"},"components":[{"name":"Multicluster Platform"}]}`
 
 **Bug required fields** (enforced by the create validator; note `createmeta` omits `versions` but the API rejects without it):
 

home/claude/memory/wsp_32957_pim_migration.md

@@ -0,0 +1,58 @@
+---
+name: wsp-32957-pim-migration
+description: State of the WSP AKS-RBAC-to-PIM migration and AutoPerm decommission, and how to resume it
+metadata:
+  node_type: memory
+  type: project
+---
+
+Long-running epic (August 2026) moving WSP's AKS cluster RBAC off **AutoPerm
+Manager**-maintained `wsp-*` groups onto the per-subscription **`CEO-*` Azure PIM**
+groups, then retiring AutoPerm. Epic **WSP-32957**; on the critical path for Zensar
+L1 on-call (WSP-32193). Work happens in **`~/code/multicluster`**
+(`terraform/cluster`, `terraform/cluster-k8s-resources`, `products/*/environments/*`).
+
+**Resume from `~/code/WSP-32957-CONTINUATION.md`** — full state, branch list,
+verified object IDs, findings and next steps. Jira is the durable record; that file
+is the index. Keep it updated as work lands ([[docs-keep-updated]]).
+
+**Landed:** WSP-33141 (multicluster PR #1808, merged `b047163c`) added object-ID
+inputs — `admin_group_oids` on `cluster`, and `cluster_user_group_oids` /
+`cluster_viewer_group_oids` / `cluster_superuser_group_oids` on
+`cluster-k8s-resources`. Supplying IDs _replaces_ the display-name lookup and leaves
+the `data "azuread_group"` unread, which is what will let the legacy groups be
+deleted. Nothing sets them yet, so behaviour is unchanged. Repoint branches for test
+(WSP-33067) and staging (WSP-33068) are pushed but have **no PR** — both gated on
+decisions, not code.
+
+**Facts that cost real effort to establish, do not re-derive:**
+
+- **Production runs in `fc7af6ae-…` (_Workspace Platform Technical Preview_), not
+  `d6d75d07-…` (_Workspace Platform Production_)**, which holds no clusters. The
+  epic was wrong about this for its whole life and every production `CEO-*` group
+  name and object ID had to change. Because `CEO-*` names embed the subscription
+  name, **always re-verify object IDs against live Entra rather than trusting the
+  epic table.**
+- **Three tenants**, not two: `6f4fe054` (prod, prod JP), `335836de` (staging,
+  staging JP, test), `3eae2746` (dev). Each `wsp-*` name is a distinct object in
+  each tenant.
+- `wsp-staging-cluster-admins` and `wsp-test-cluster-admins` are **empty**, so
+  `wsp-owner` is the _only_ path to `cluster-admin` in staging. Never drop it before
+  `SuperAdmin-*` is proven — hence the staging branch is split into an additive
+  commit and a cutover commit.
+- **No break-glass exists**: `disableLocalAccounts = True` on every cluster; only the
+  deployment SPNs authenticate non-interactively.
+- Graph **PIM-for-Groups is unreadable via `az`** (the CLI's first-party client lacks
+  `PrivilegedAccess.Read.AzureADGroup`, on both `v1.0` and `beta`). Use the portal or
+  `Connect-MgGraph -Scopes PrivilegedAccess.Read.AzureADGroup`.
+
+**Watch for:** the epic gets rewritten by James Weldrake between sessions — re-read
+the description before acting, and check which child tickets are still live
+(WSP-33062/33063/33064/33066 were cancelled 2026-08-24, and dev was put out of
+scope). Verified findings have repeatedly contradicted the epic text
+([[copilot-review-false-positives]] is the same instinct: check against reality
+first).
+
+Queued Slack messages and the leaver report live as `~/code/*.txt` alongside the
+continuation file; see the table in it for what has and has not been sent
+([[workflow-review-and-comments]] — show them before they go out).

home/shell.nix

@@ -287,7 +287,22 @@ in
     plugins = with pkgs.tmuxPlugins; [
       sensible
       vim-tmux-navigator # Ctrl-h/j/k/l across vim splits and tmux panes
-      yank
+      {
+        # On WSL, tmux-yank pipes the selection to clip.exe, which decodes its
+        # stdin as the OEM codepage instead of UTF-8 -- an em dash reaches the
+        # Windows clipboard as three characters. Route through tmux's own
+        # buffer instead: with set-clipboard on, tmux emits OSC 52 and the
+        # terminal takes the text as UTF-8. Windows Terminal honours OSC 52;
+        # iTerm2 does not by default, hence the runtime guard rather than
+        # overriding pbcopy/xsel on every host. yank.tmux bakes the command
+        # into its key bindings when it loads, so this must be set first, which
+        # is what plugin extraConfig gives us.
+        plugin = yank;
+        extraConfig = ''
+          if-shell 'grep -qi microsoft /proc/version 2>/dev/null' \
+            "set -g @override_copy_command 'tmux load-buffer -w -'"
+        '';
+      }
       extrakto # prefix+Tab: fzf-grab paths/URLs/text from the pane into the prompt
       {
         # Catppuccin Mocha statusline (v2 API: flavour + window options must be
@@ -298,6 +313,14 @@ in
         extraConfig = ''
           set -g @catppuccin_flavor 'mocha'
           set -g @catppuccin_window_status_style 'rounded'
+          # Catppuccin's default window text is #T, the pane title, which every
+          # program in the pane is free to overwrite -- the shell writes the
+          # hostname, Claude Code writes its current task, and a hand-set window
+          # name never appears. Show the window name instead, falling back to the
+          # pane title when the window holds a single pane and the two carry the
+          # same information anyway.
+          set -g @catppuccin_window_text ' #{?#{==:#{window_panes},1},#T,#W}'
+          set -g @catppuccin_window_current_text ' #{?#{==:#{window_panes},1},#T,#W}'
         '';
       }
       resurrect # save/restore sessions
@@ -342,6 +365,17 @@ in
       set -g renumber-windows on
       set -g set-clipboard on
 
+      # Pane titles on the border, but only once a window is split -- a single
+      # pane's title is already in the status bar. pane-border-status takes no
+      # format, so the hook recomputes it whenever the layout changes, which
+      # covers both splitting and closing a pane.
+      set -g pane-border-format " #P #{pane_title} "
+      set -g pane-border-status off
+      set-hook -g window-layout-changed 'set -Fw pane-border-status "#{?#{>:#{window_panes},1},top,off}"'
+
+      # Pane titles have no default binding.
+      bind T command-prompt -p "pane title:" "select-pane -T '%%'"
+
       # Catppuccin v2 statusline. Must run after the plugin has loaded;
       # home-manager appends this extraConfig after the whole plugin list.
       set -g status-left-length 100

hosts/Darwin/configuration.nix

@@ -161,6 +161,11 @@
     dock = {
       show-recents = false;
       mru-spaces = false; # don't reorder spaces by use
+      # Disable hot-corners
+      wvous-tr-corner = 1;
+      wvous-tl-corner = 1;
+      wvous-bl-corner = 1;
+      wvous-br-corner = 1;
     };
     finder = {
       AppleShowAllExtensions = true;

hosts/EDaaS/configuration.nix

@@ -60,6 +60,11 @@
   ## patch the script
   systemd.services.docker-desktop-proxy.script = lib.mkForce ''${config.wsl.wslConf.automount.root}/wsl/docker-desktop/docker-desktop-user-distro proxy --docker-desktop-root ${config.wsl.wslConf.automount.root}/wsl/docker-desktop "C:\Program Files\Docker\Docker\resources"'';
 
+  # NixOS-WSL's passwordless wheel default only covers `security.sudo`; the
+  # sudo-rs swap in common-nixos.nix needs it set again. No console login here,
+  # and no account password anyone knows.
+  security.sudo-rs.wheelNeedsPassword = false;
+
   features.swayDesktop.enable = false;
 
   # NOTE: this user's systemd --user lingering -- so the home-manager renovate

hosts/RPi5/configuration.nix

@@ -17,7 +17,7 @@
 
   # Headless server: the Sway desktop is intentionally not set up. modules/sway.nix is
   # not imported and features.swayDesktop.enable defaults to false (declared in
-  # system/modules/features.nix), so this host keeps plain TTY/SSH login.
+  # modules/features.nix), so this host keeps plain TTY/SSH login.
 
   # Raspberry Pi boots via U-Boot + extlinux, not GRUB/systemd-boot. The
   # raspberry-pi-5 nixos-hardware profile supplies the kernel, firmware and

modules/sway.nix

@@ -12,7 +12,7 @@ let
 in
 {
   # The features.swayDesktop.enable option is declared in
-  # system/modules/features.nix (so headless hosts can read/set it without
+  # modules/features.nix (so headless hosts can read/set it without
   # importing this module). This module only provides its implementation.
   config = lib.mkIf cfg.enable {
     programs.sway = {

modules/users.nix

@@ -29,10 +29,10 @@
     // lib.optionalAttrs (spec ? linger) { inherit (spec) linger; }
   ) hostUsers;
 
-  programs.firefox = lib.mkIf (config.features.swayDesktop.enable == true) {
+  programs.firefox = lib.mkIf config.features.swayDesktop.enable {
     enable = true;
   };
-  programs.thunderbird = lib.mkIf (config.features.swayDesktop.enable == true) {
+  programs.thunderbird = lib.mkIf config.features.swayDesktop.enable {
     enable = true;
   };
 }

users/emmathorpe/work.nix

@@ -47,7 +47,23 @@
     pkgs.terraform-docs # generate Terraform module docs
     pkgs.yq-go # jq for YAML
     pkgs.gcx # Grafana Cloud CLI (dashboards, SLOs, synthetics, alerts)
+
+    # WSL ships no xdg-open, so anything that shells out to a browser dies with
+    # `exec: "xdg-open,x-www-browser,www-browser": executable file not found`.
+    # kubelogin's interactive login is the one that bites: it is the login mode
+    # the shared cluster kubeconfig uses. Hand the URL to Windows instead.
+    # (wslu, the usual answer, is gone from nixpkgs -- upstream archived it.)
+    (pkgs.writeShellScriptBin "xdg-open" ''
+      url="$1"
+      if command -v powershell.exe >/dev/null 2>&1; then
+        exec powershell.exe -NoProfile -Command "Start-Process '$url'"
+      fi
+      exec explorer.exe "$url"
+    '')
   ];
+
+  # Honoured by tools that read $BROWSER rather than calling xdg-open.
+  home.sessionVariables.BROWSER = "xdg-open";
   services.ssh-agent.enable = true;
 
   # Colourised kubectl. enableAlias points `kubectl` at kubecolor, which parses