Merge remote-tracking branch 'origin/main' into fix-ssh-authorized-users-multi-rule

# Conflicts:
#	management/server/types/account_components.go
#	shared/management/types/networkmap_components.go

.devcontainer/Dockerfile

@@ -1,4 +1,4 @@
-FROM golang:1.25-bookworm
+FROM golang:1.26.7-bookworm
 
 RUN apt-get update && export DEBIAN_FRONTEND=noninteractive \
     && apt-get -y install --no-install-recommends\

.git-branches.toml

@@ -3,7 +3,7 @@
 [branches]
 main = "main"
 perennials = []
-perennial-regex = ""
+perennial-regex = "^release-"
 
 [create]
 new-branch-type = "feature"

.github/workflows/agent-network-e2e.yml

@@ -12,6 +12,13 @@ on:
           AWS issues it. Leave empty for the Sonnet 4.6 default.
         required: false
         default: ""
+      test_pattern:
+        description: >-
+          Package pattern to run. Defaults to the whole suite; narrow it to one
+          package (e.g. ./e2e/agentnetwork/...) when a run only needs that
+          package's answer and not the sixteen minutes the container suite costs.
+        required: false
+        default: "./e2e/..."
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
@@ -77,4 +84,8 @@ jobs:
           GOOGLE_VERTEX_PROJECT: ${{ secrets.E2E_GOOGLE_VERTEX_PROJECT }}
           GOOGLE_VERTEX_REGION: ${{ secrets.E2E_GOOGLE_VERTEX_REGION }}
           GOOGLE_VERTEX_MODEL: ${{ secrets.E2E_GOOGLE_VERTEX_MODEL }}
-        run: go test -tags e2e -timeout 40m -v ./e2e/...
+          # Read through an env var rather than interpolated into the run
+          # script: a dispatch input reaching a shell command directly is a
+          # script-injection seam, however trusted the dispatcher.
+          TEST_PATTERN: ${{ inputs.test_pattern || './e2e/...' }}
+        run: go test -tags e2e -timeout 40m -v "$TEST_PATTERN"

.github/workflows/buf.yml

@@ -0,0 +1,33 @@
+name: protobuf checks
+on:
+  push:
+    branches:
+      - main
+      - "release-*"
+  pull_request:
+    paths:
+      - ".github/workflows/buf.yml"
+      - "**/buf.yaml"
+      - "**/buf.lock"
+      - "**/buf.gen.yaml"
+      - "**.proto"
+permissions:
+  contents: read
+  pull-requests: read
+jobs:
+  buf:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        with:
+          persist-credentials: false
+      - uses: bufbuild/buf-action@8c6a16e16f12ba20b6470afa9c2ba9b5ba8c97c3 # v1.5.0
+        with:
+          push: false
+          archive: false
+          pr_comment: false
+          build: false
+          lint: false
+          format: false
+          breaking: true

.github/workflows/check-license-dependencies.yml

@@ -2,7 +2,7 @@ name: Check License Dependencies
 
 on:
   push:
-    branches: [main]
+    branches: [main, "release-*"]
     paths:
       - "go.mod"
       - "go.sum"

.github/workflows/frontend-ui.yml

@@ -10,6 +10,7 @@ on:
   push:
     branches:
       - main
+      - "release-*"
     paths:
       - "client/ui/frontend/**"
       - "client/ui/i18n/**"

.github/workflows/golang-test-darwin.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - main
+      - "release-*"
   pull_request:
 
 concurrency:

.github/workflows/golang-test-freebsd.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - main
+      - "release-*"
   pull_request:
 
 concurrency:

.github/workflows/golang-test-linux.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - main
+      - "release-*"
   pull_request:
 
 concurrency:
@@ -232,7 +233,7 @@ jobs:
             -e GOCACHE=${CONTAINER_GOCACHE} \
             -e GOMODCACHE=${CONTAINER_GOMODCACHE} \
             -e CONTAINER=${CONTAINER} \
-            golang:1.25-alpine \
+            golang:1.26.7-alpine \
             sh -c ' \
               apk update; apk add --no-cache \
                 ca-certificates iptables ip6tables dbus dbus-dev libpcap-dev build-base; \
@@ -729,6 +730,11 @@ jobs:
       - name: Install modules
         run: go mod tidy
 
+      - name: Run Mage
+        uses: magefile/mage-action@a662bd8c29d8106879588cfff83b2faf6e6f59db # v4.0.0
+        with:
+          install-only: true
+
       - name: check git status
         run: git --no-pager diff --exit-code
 
@@ -737,9 +743,7 @@ jobs:
           CGO_ENABLED=1 GOARCH=${{ matrix.arch }} \
           NETBIRD_STORE_ENGINE=${{ matrix.store }} \
           CI=true \
-          go test -tags=integration -coverprofile=coverage.txt \
-          -exec 'sudo --preserve-env=CI,NETBIRD_STORE_ENGINE' \
-          -timeout 20m ./management/server/http/...
+          mage integrationtest:all -gotestflags="-coverprofile=coverage.txt"
 
       - name: Upload coverage reports to Codecov
         if: matrix.arch == 'amd64'

.github/workflows/golang-test-windows.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - main
+      - "release-*"
   pull_request:
 
 env:

.github/workflows/install-script-test.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - main
+      - "release-*"
   pull_request:
     paths:
       - "release_files/install.sh"

.github/workflows/mobile-build-validation.yml

@@ -1,71 +0,0 @@
-name: Mobile
-
-on:
-  push:
-    branches:
-      - main
-  pull_request:
-
-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.actor_id }}
-  cancel-in-progress: true
-
-jobs:
-  android_build:
-    name: "Android / Build"
-    runs-on: ubuntu-latest
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
-        with:
-          persist-credentials: false
-      - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
-        with:
-          go-version-file: "go.mod"
-      - name: Setup Android SDK
-        uses: android-actions/setup-android@40fd30fb8d7440372e1316f5d1809ec01dcd3699 # v4.0.1
-        with:
-          cmdline-tools-version: 8512546
-      - name: Setup Java
-        uses: actions/setup-java@1bcf9fb12cf4aa7d266a90ae39939e61372fe520
-        with:
-          java-version: "11"
-          distribution: "adopt"
-      - name: NDK Cache
-        id: ndk-cache
-        uses: actions/cache@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
-        with:
-          path: /usr/local/lib/android/sdk/ndk
-          key: ndk-cache-23.1.7779620
-      - name: Setup NDK
-        run: /usr/local/lib/android/sdk/cmdline-tools/7.0/bin/sdkmanager --install "ndk;23.1.7779620"
-      - name: install gomobile
-        run: go install golang.org/x/mobile/cmd/gomobile@v0.0.0-20251113184115-a159579294ab
-      - name: gomobile init
-        run: gomobile init
-      - name: build android netbird lib
-        run: PATH=$PATH:$(go env GOPATH) gomobile bind -o $GITHUB_WORKSPACE/netbird.aar -javapkg=io.netbird.gomobile -ldflags="-checklinkname=0 -X golang.zx2c4.com/wireguard/ipc.socketDirectory=/data/data/io.netbird.client/cache/wireguard -X github.com/netbirdio/netbird/version.version=buildtest"  $GITHUB_WORKSPACE/client/android
-        env:
-          CGO_ENABLED: 0
-          ANDROID_NDK_HOME: /usr/local/lib/android/sdk/ndk/23.1.7779620
-  ios_build:
-    name: "iOS / Build"
-    runs-on: macos-latest
-    steps:
-      - name: Checkout repository
-        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
-        with:
-          persist-credentials: false
-      - name: Install Go
-        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
-        with:
-          go-version-file: "go.mod"
-      - name: install gomobile
-        run: go install golang.org/x/mobile/cmd/gomobile@v0.0.0-20251113184115-a159579294ab
-      - name: gomobile init
-        run: gomobile init
-      - name: build iOS netbird lib
-        run: PATH=$PATH:$(go env GOPATH) gomobile bind -target=ios -bundleid=io.netbird.framework -ldflags="-X github.com/netbirdio/netbird/version.version=buildtest" -o ./NetBirdSDK.xcframework ./client/ios/NetBirdSDK
-        env:
-          CGO_ENABLED: 0

.github/workflows/no-new-replace.yml

@@ -0,0 +1,78 @@
+name: No New Replace Directives
+
+on:
+  pull_request:
+    paths:
+      - "go.mod"
+
+permissions:
+  contents: read
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.actor_id }}
+  cancel-in-progress: true
+
+jobs:
+  check-replace-directives:
+    name: check-replace-directives
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        with:
+          persist-credentials: false
+          fetch-depth: 0
+
+      - name: Install Go
+        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        with:
+          go-version-file: go.mod
+
+      - name: Compare replace directives against the base branch
+        env:
+          BASE_SHA: ${{ github.event.pull_request.base.sha }}
+        run: |
+          set -euo pipefail
+
+          # A replace directive only applies when this module is the main
+          # module. Anything importing netbird as a library, the embedded
+          # clients among them, resolves the replaced path upstream instead and
+          # fails to build against whatever the replacement provides. Requiring
+          # a fork under its own module path avoids that; a replace does not.
+          #
+          # go.mod is parsed rather than diffed so that reordering, comments and
+          # single-line versus block syntax do not register as changes.
+          #
+          # Versions are part of the key because a replace can be scoped to one
+          # version of a module. Keyed on paths alone, retargeting such a
+          # directive at a different version would read as unchanged.
+          list_replaces() {
+            go mod edit -json "$1" \
+              | jq -r '
+                  def ref: .Path + (if (.Version // "") == "" then "" else " " + .Version end);
+                  (.Replace // [])[] | "\(.Old | ref) => \(.New | ref)"
+                ' \
+              | sort
+          }
+
+          git show "${BASE_SHA}:go.mod" > /tmp/base-go.mod
+          list_replaces /tmp/base-go.mod > /tmp/base-replaces
+          list_replaces go.mod > /tmp/head-replaces
+
+          added=$(comm -13 /tmp/base-replaces /tmp/head-replaces)
+          if [ -n "$added" ]; then
+            echo "::error::This PR adds a replace directive to go.mod:"
+            echo "$added" | sed 's/^/  /'
+            echo ""
+            echo "A replace directive applies only to the main module, so it does not"
+            echo "reach anything that imports netbird as a library. Require the module"
+            echo "under a path you control instead, as done for github.com/netbirdio/go-nat."
+            exit 1
+          fi
+
+          removed=$(comm -23 /tmp/base-replaces /tmp/head-replaces)
+          if [ -n "$removed" ]; then
+            echo "This PR removes replace directives:"
+            echo "$removed" | sed 's/^/  /'
+          fi
+          echo "No new replace directives."

.github/workflows/pr-title-check.yml

@@ -16,6 +16,8 @@ jobs:
             const allowedTags = [
               'management',
               'client',
+              'android',
+              'ios',
               'signal',
               'proxy',
               'relay',

.github/workflows/release.yml

@@ -6,6 +6,7 @@ on:
       - "v*"
     branches:
       - main
+      - "release-*"
   pull_request:
 
 env:
@@ -214,7 +215,7 @@ jobs:
           echo "GPG_RPM_KEY_FILE=/tmp/gpg-rpm-signing-key.asc" >> $GITHUB_ENV
 
       - name: Install goversioninfo
-        run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@233067e
+        run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@b66839b # v1.7.0
       - name: Generate windows syso amd64
         run: goversioninfo  -icon client/ui/build/windows/icon.ico -manifest client/manifest.xml -product-name ${{ env.PRODUCT_NAME }} -copyright "${{ env.COPYRIGHT }}" -ver-major ${{ steps.semver_parser.outputs.major }} -ver-minor ${{ steps.semver_parser.outputs.minor }} -ver-patch ${{ steps.semver_parser.outputs.patch }} -ver-build 0 -file-version ${{ steps.semver_parser.outputs.fullversion }}.0 -product-version ${{ steps.semver_parser.outputs.fullversion }}.0 -o client/resources_windows_amd64.syso
       - name: Generate windows syso arm64
@@ -254,15 +255,23 @@ jobs:
         id: tag_and_push_images
         if: |
           (github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository) ||
-          (github.event_name == 'push' && github.ref == 'refs/heads/main')
+          (github.event_name == 'push' && (github.ref == 'refs/heads/main' || startsWith(github.ref, 'refs/heads/release-')))
         run: |
           set -euo pipefail
 
+          # $GITHUB_REF / $GITHUB_EVENT_NAME are read from the runner
+          # environment rather than substituted into this script with the
+          # workflow expression syntax: branch names may legally contain
+          # $(…), and interpolating github.ref would execute it.
           resolve_tags() {
-            if [[ "${{ github.event_name }}" == "pull_request" ]]; then
+            if [[ "$GITHUB_EVENT_NAME" == "pull_request" ]]; then
               echo "pr-${{ github.event.pull_request.number }}"
-            else
+            elif [[ "$GITHUB_REF" == "refs/heads/main" ]]; then
               echo "main sha-$(git rev-parse --short HEAD)"
+            else
+              # Release branches get an immutable sha-* tag only — the floating
+              # "main" tag must never move from a release branch.
+              echo "sha-$(git rev-parse --short HEAD)"
             fi
           }
 
@@ -426,7 +435,7 @@ jobs:
           tar -xf llvm-mingw-20250709-ucrt-ubuntu-22.04-x86_64.tar.xz
           echo "/tmp/llvm-mingw-20250709-ucrt-ubuntu-22.04-x86_64/bin" >> $GITHUB_PATH
       - name: Install goversioninfo
-        run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@233067e
+        run: go install github.com/josephspurrier/goversioninfo/cmd/goversioninfo@b66839b # v1.7.0
       - name: Install wails3 CLI
         # Version derived from go.mod so the binding generator always matches
         # the wails runtime the binary links against.
@@ -475,6 +484,132 @@ jobs:
           path: dist/
           retention-days: 3
 
+  release_ui_gtk3:
+    # Legacy GTK3/WebKit2GTK 4.1 UI build for distros without WebKitGTK 6.0
+    # (Ubuntu 22.04, Debian 12, RHEL 9, Fedora <=39). Runs on ubuntu-22.04 so
+    # the binary links against the oldest supported glibc.
+    runs-on: ubuntu-22.04
+    outputs:
+      release_ui_gtk3_artifact_url: ${{ steps.upload_release_ui_gtk3.outputs.artifact-url }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@9c091bb21b7c1c1d1991bb908d89e4e9dddfe3e0 # v7.0.0
+        with:
+          fetch-depth: 0 # It is required for GoReleaser to work properly
+          persist-credentials: false
+
+      - name: Parse semver string
+        id: semver_parser
+        uses: netbirdio/shared-actions/actions/parse-semver@be5df6047383da2236e02243cceb857d8567c27e # v0.0.2
+
+      - name: Set snapshot flag
+        if: ${{ !startsWith(github.ref, 'refs/tags/v') }}
+        run: |
+          echo "flags=--snapshot" >> $GITHUB_ENV
+
+      - name: Set build vars
+        if: ${{ startsWith(github.ref, 'refs/tags/v') }}
+        run: |
+          if [[ "x-${{ steps.semver_parser.outputs.prerelease }}" == "x-" && "x-${{ github.repository }}" == "x-netbirdio/netbird" ]]; then
+            echo "x-${{ github.repository }}"
+            echo "x-${{ steps.semver_parser.outputs.prerelease }}"
+            echo "SKIP_PUBLISH=false" >> $GITHUB_ENV
+          else
+            echo "x-${{ github.repository }}"
+            echo "x-${{ steps.semver_parser.outputs.prerelease }}"
+          fi
+
+      - name: Set up Go
+        uses: actions/setup-go@924ae3a1cded613372ab5595356fb5720e22ba16 # v6.5.0
+        with:
+          go-version-file: "go.mod"
+          cache: false
+      - name: Cache Go modules
+        # Restore-only from the release_ui cache written by trusted runs; the
+        # module cache is identical (same go.sum) and stale build-cache
+        # entries just miss.
+        uses: actions/cache/restore@2c8a9bd7457de244a408f35966fab2fb45fda9c8 # v6.0.0
+        with:
+          path: |
+            ~/go/pkg/mod
+            ~/.cache/go-build
+          key: ${{ runner.os }}-ui-go-releaser-${{ hashFiles('**/go.sum') }}
+          restore-keys: |
+            ${{ runner.os }}-ui-go-releaser-
+
+      - name: Install modules
+        run: go mod tidy
+
+      - name: check git status
+        run: git --no-pager diff --exit-code
+
+      - name: Set up Node.js
+        uses: actions/setup-node@49933ea5288caeca8642d1e84afbd3f7d6820020 # v4
+        with:
+          node-version: '22'
+
+      - name: Set up pnpm
+        uses: pnpm/action-setup@a3252b78c470c02df07e9d59298aecedc3ccdd6d # v3.0.0
+        with:
+          version: 11
+
+      - name: Install dependencies
+        run: sudo apt update && sudo apt install -y -q libgtk-3-dev libwebkit2gtk-4.1-dev
+
+      - name: Decode GPG signing key
+        if: github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository
+        env:
+          GPG_RPM_PRIVATE_KEY: ${{ secrets.GPG_RPM_PRIVATE_KEY }}
+        run: |
+          echo "$GPG_RPM_PRIVATE_KEY" | base64 -d > /tmp/gpg-rpm-signing-key.asc
+          echo "GPG_RPM_KEY_FILE=/tmp/gpg-rpm-signing-key.asc" >> $GITHUB_ENV
+
+      - name: Install wails3 CLI
+        # Version derived from go.mod so the binding generator always matches
+        # the wails runtime the binary links against.
+        # -tags gtk3: the CLI links the wails runtime's cgo packages, and the
+        # default tags request gtk4/webkitgtk-6.0 pkg-config entries that do
+        # not exist on ubuntu-22.04.
+        run: |
+          WAILS_VERSION=$(go list -m -f '{{.Version}}' github.com/wailsapp/wails/v3)
+          go install -tags gtk3 github.com/wailsapp/wails/v3/cmd/wails3@$WAILS_VERSION
+
+      - name: Run GoReleaser
+        uses: goreleaser/goreleaser-action@5daf1e915a5f0af01ddbcd89a43b8061ff4f1a89 # v7.2.2
+        with:
+          version: ${{ env.GORELEASER_VER }}
+          args: release --config .goreleaser_ui_gtk3.yaml --clean ${{ env.flags }}
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+          UPLOAD_DEBIAN_SECRET: ${{ secrets.PKG_UPLOAD_SECRET }}
+          UPLOAD_YUM_SECRET: ${{ secrets.PKG_UPLOAD_SECRET }}
+          GPG_RPM_KEY_FILE: ${{ env.GPG_RPM_KEY_FILE }}
+          NFPM_NETBIRD_UI_RPM_GTK3_PASSPHRASE: ${{ secrets.GPG_RPM_PASSPHRASE }}
+          SKIP_PUBLISH: ${{ env.SKIP_PUBLISH }}
+      - name: Verify RPM signatures
+        run: |
+          docker run --rm -v $(pwd)/dist:/dist fedora:41 bash -c '
+            dnf install -y -q rpm-sign curl >/dev/null 2>&1
+            curl -sSL https://pkgs.netbird.io/yum/repodata/repomd.xml.key -o /tmp/rpm-pub.key
+            rpm --import /tmp/rpm-pub.key
+            echo "=== Verifying RPM signatures ==="
+            for rpm_file in /dist/*.rpm; do
+              [ -f "$rpm_file" ] || continue
+              echo "--- $(basename $rpm_file) ---"
+              rpm -K "$rpm_file"
+            done
+          '
+      - name: Clean up GPG key
+        if: always()
+        run: rm -f /tmp/gpg-rpm-signing-key.asc
+      - name: upload non tags for debug purposes
+        id: upload_release_ui_gtk3
+        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a #v7.0.1
+        with:
+          name: release-ui-gtk3
+          path: dist/
+          retention-days: 3
+
   release_ui_darwin:
     runs-on: macos-latest
     outputs:
@@ -688,7 +823,7 @@ jobs:
   comment_release_artifacts:
     name: Comment release artifacts
     runs-on: ubuntu-latest
-    needs: [release, release_ui, release_ui_darwin]
+    needs: [release, release_ui, release_ui_gtk3, release_ui_darwin]
     if: ${{ always() && github.event_name == 'pull_request' && github.event.pull_request.head.repo.full_name == github.repository }}
     permissions:
       contents: read
@@ -700,12 +835,14 @@ jobs:
         env:
           RELEASE_RESULT: ${{ needs.release.result }}
           RELEASE_UI_RESULT: ${{ needs.release_ui.result }}
+          RELEASE_UI_GTK3_RESULT: ${{ needs.release_ui_gtk3.result }}
           RELEASE_UI_DARWIN_RESULT: ${{ needs.release_ui_darwin.result }}
           RELEASE_ARTIFACT_URL: ${{ needs.release.outputs.release_artifact_url }}
           LINUX_PACKAGES_ARTIFACT_URL: ${{ needs.release.outputs.linux_packages_artifact_url }}
           WINDOWS_PACKAGES_ARTIFACT_URL: ${{ needs.release.outputs.windows_packages_artifact_url }}
           MACOS_PACKAGES_ARTIFACT_URL: ${{ needs.release.outputs.macos_packages_artifact_url }}
           RELEASE_UI_ARTIFACT_URL: ${{ needs.release_ui.outputs.release_ui_artifact_url }}
+          RELEASE_UI_GTK3_ARTIFACT_URL: ${{ needs.release_ui_gtk3.outputs.release_ui_gtk3_artifact_url }}
           RELEASE_UI_DARWIN_ARTIFACT_URL: ${{ needs.release_ui_darwin.outputs.release_ui_darwin_artifact_url }}
           GHCR_IMAGES_MARKDOWN: ${{ needs.release.outputs.ghcr_images }}
         with:
@@ -728,6 +865,7 @@ jobs:
               ['Windows packages', process.env.WINDOWS_PACKAGES_ARTIFACT_URL, process.env.RELEASE_RESULT],
               ['macOS packages', process.env.MACOS_PACKAGES_ARTIFACT_URL, process.env.RELEASE_RESULT],
               ['UI artifacts', process.env.RELEASE_UI_ARTIFACT_URL, process.env.RELEASE_UI_RESULT],
+              ['UI GTK3 artifacts', process.env.RELEASE_UI_GTK3_ARTIFACT_URL, process.env.RELEASE_UI_GTK3_RESULT],
               ['UI macOS artifacts', process.env.RELEASE_UI_DARWIN_ARTIFACT_URL, process.env.RELEASE_UI_DARWIN_RESULT],
             ];
 
@@ -784,7 +922,7 @@ jobs:
 
   trigger_signer:
     runs-on: ubuntu-latest
-    needs: [release, release_ui, release_ui_darwin, test_windows_installer]
+    needs: [release, release_ui, release_ui_gtk3, release_ui_darwin, test_windows_installer]
     if: startsWith(github.ref, 'refs/tags/')
     steps:
       - name: Trigger binaries sign pipelines

.github/workflows/sync-tag.yml

@@ -9,21 +9,9 @@ concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.actor_id }}
   cancel-in-progress: true
 
-# Receiving workflows (cloud sync-tag, mobile bump-netbird) expect the short
-# tag form (e.g. v0.30.0), not refs/tags/v0.30.0 — github.ref_name, not github.ref.
+# The receiving bump-netbird workflows expect the short tag form
+# (e.g. v0.30.0), not refs/tags/v0.30.0 — github.ref_name, not github.ref.
 jobs:
-  trigger_sync_tag:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Trigger release tag sync
-        uses: benc-uk/workflow-dispatch@31e2b3319479a63f0ab15bf800eff9e913504e26 # v1.3.2
-        with:
-          workflow: sync-tag.yml
-          ref: main
-          repo: ${{ secrets.UPSTREAM_REPO }}
-          token: ${{ secrets.NC_GITHUB_TOKEN }}
-          inputs: '{ "tag": "${{ github.ref_name }}" }'
-
   trigger_android_bump:
     runs-on: ubuntu-latest
     if: github.event.created && !github.event.deleted && startsWith(github.ref, 'refs/tags/v') && !contains(github.ref_name, '-')
@@ -49,3 +37,16 @@ jobs:
           repo: netbirdio/ios-client
           token: ${{ secrets.NC_GITHUB_TOKEN }}
           inputs: '{ "tag": "${{ github.ref_name }}" }'
+
+  trigger_dashboard_bump:
+    runs-on: ubuntu-latest
+    if: github.event.created && !github.event.deleted && startsWith(github.ref, 'refs/tags/v') && !contains(github.ref_name, '-')
+    steps:
+      - name: Trigger dashboard wasm client bump
+        uses: benc-uk/workflow-dispatch@31e2b3319479a63f0ab15bf800eff9e913504e26 # v1.3.2
+        with:
+          workflow: bump-netbird.yml
+          ref: main
+          repo: netbirdio/dashboard
+          token: ${{ secrets.NC_GITHUB_TOKEN }}
+          inputs: '{ "tag": "${{ github.ref_name }}" }'

.github/workflows/test-infrastructure-files.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - main
+      - "release-*"
   pull_request:
     paths:
       - "infrastructure_files/**"
@@ -257,6 +258,15 @@ jobs:
         with:
           persist-credentials: false
 
+      - name: Verify fresh-install session cookie key hardening
+        run: |
+          grep -Fxq '  SESSION_COOKIE_ENCRYPTION_KEY=$(openssl rand -base64 32)' infrastructure_files/getting-started.sh
+          grep -Fxq '    sessionCookieEncryptionKey: "$SESSION_COOKIE_ENCRYPTION_KEY"' infrastructure_files/getting-started.sh
+          grep -Fxq '  install -m 600 /dev/null config.yaml' infrastructure_files/getting-started.sh
+          grep -Fxq '  openssl rand -base64 32' infrastructure_files/getting-started-enterprise.sh
+          grep -Fxq '  NETBIRD_SESSION_COOKIE_ENCRYPTION_KEY=$(rand_b64_key)' infrastructure_files/getting-started-enterprise.sh
+          grep -Fxq '    sessionCookieEncryptionKey: "${NETBIRD_SESSION_COOKIE_ENCRYPTION_KEY}"' infrastructure_files/getting-started-enterprise.sh
+
       - name: Verify Dex retirement notice
         run: |
           if infrastructure_files/getting-started-with-dex.sh >stdout.txt 2>stderr.txt; then

.github/workflows/ui-translations.yml

@@ -0,0 +1,42 @@
+name: UI Translations
+
+on:
+  pull_request:
+    paths:
+      - "client/ui/i18n/locales/**"
+      - "client/ui/i18n/check-translations.mjs"
+      - ".github/workflows/ui-translations.yml"
+  push:
+    branches:
+      - main
+    paths:
+      - "client/ui/i18n/locales/**"
+      - "client/ui/i18n/check-translations.mjs"
+
+permissions:
+  contents: read
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.actor_id }}
+  cancel-in-progress: true
+
+jobs:
+  check-translations:
+    name: Check translation key parity
+    runs-on: ubuntu-latest
+    timeout-minutes: 5
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2
+        with:
+          persist-credentials: false
+
+      - name: Set up Node.js
+        uses: actions/setup-node@v4
+        with:
+          node-version: "22"
+
+      # English (en) is the source of truth for translation keys; every other
+      # locale declared in _index.json must carry the exact same key set.
+      - name: Check translation key parity
+        run: node client/ui/i18n/check-translations.mjs

.github/workflows/wasm-build-validation.yml

@@ -4,6 +4,7 @@ on:
   push:
     branches:
       - main
+      - "release-*"
   pull_request:
 
 concurrency:

.goreleaser.yaml

@@ -468,6 +468,13 @@ checksum:
     - glob: ./infrastructure_files/migrate-to-enterprise.sh
 
 release:
+  # The signing pipeline (netbirdio/sign-pipelines, dispatched by
+  # trigger_signer) marks the release latest once the Windows and macOS
+  # artifacts are signed. Without this override goreleaser marks it latest
+  # at publish time, while those artifacts are still unsigned.
+  make_latest: false
+  # Mark x.y.z-rc.* and other prerelease tags as prereleases on GitHub.
+  prerelease: auto
   extra_files:
     - glob: ./infrastructure_files/getting-started-with-zitadel.sh
     - glob: ./release_files/install.sh

.goreleaser_ui.yaml

@@ -92,8 +92,16 @@ nfpms:
         dst: /usr/share/applications/org.wails.netbird.desktop
       - src: client/ui/build/appicon.png
         dst: /usr/share/pixmaps/netbird.png
+      # Names the polkit action for the elevation prompt the app raises when an
+      # unprivileged user changes a privileged setting; without it the dialog
+      # shows a raw command line.
+      - src: client/ui/build/linux/polkit/io.netbird.settings.policy
+        dst: /usr/share/polkit-1/actions/io.netbird.settings.policy
     dependencies:
-      - netbird
+      - netbird (>= 0.75.0)
+      - libgtk-4-1 (>= 4.14)
+      - libwebkitgtk-6.0-4
+      - xdg-utils
 
   - maintainer: Netbird <dev@netbird.io>
     description: Netbird client UI.
@@ -113,8 +121,16 @@ nfpms:
         dst: /usr/share/applications/org.wails.netbird.desktop
       - src: client/ui/build/appicon.png
         dst: /usr/share/pixmaps/netbird.png
+      # Names the polkit action for the elevation prompt the app raises when an
+      # unprivileged user changes a privileged setting; without it the dialog
+      # shows a raw command line.
+      - src: client/ui/build/linux/polkit/io.netbird.settings.policy
+        dst: /usr/share/polkit-1/actions/io.netbird.settings.policy
     dependencies:
-      - netbird
+      - netbird >= 0.75.0
+      - (gtk4 >= 4.14 or libgtk-4-1 >= 4.14)
+      - (webkitgtk6.0 or libwebkitgtk-6_0-4)
+      - xdg-utils
 
     rpm:
       signature:
@@ -138,3 +154,11 @@ uploads:
     target: https://pkgs.wiretrustee.com/yum/{{ .Arch }}{{ if .Arm }}{{ .Arm }}{{ end }}
     username: dev@wiretrustee.com
     method: PUT
+
+release:
+  # Uploads into the release created by the main .goreleaser.yaml run.
+  # make_latest stays false everywhere: the signing pipeline
+  # (netbirdio/sign-pipelines) marks the release latest after the Windows
+  # and macOS artifacts are signed.
+  make_latest: false
+  prerelease: auto

.goreleaser_ui_darwin.yaml

@@ -43,3 +43,11 @@ checksum:
   name_template: "{{ .ProjectName }}_darwin_checksums.txt"
 changelog:
   disable: true
+
+release:
+  # Uploads into the release created by the main .goreleaser.yaml run.
+  # make_latest stays false everywhere: the signing pipeline
+  # (netbirdio/sign-pipelines) marks the release latest after the Windows
+  # and macOS artifacts are signed.
+  make_latest: false
+  prerelease: auto

.goreleaser_ui_gtk3.yaml

@@ -0,0 +1,144 @@
+version: 2
+env:
+  - SKIP_PUBLISH={{ if index .Env "SKIP_PUBLISH" }}{{ .Env.SKIP_PUBLISH }}{{ else }}true{{ end }}
+project_name: netbird-ui
+
+before:
+  hooks:
+    # Bindings are gitignored; regenerate before the frontend build so
+    # the @wailsio/runtime Vite plugin can resolve them (vite refuses to
+    # build without them).
+    # -f '-tags gtk3': the generator type-checks client/ui, whose cgo imports
+    # would otherwise resolve gtk4/webkitgtk-6.0 pkg-config entries that do
+    # not exist on ubuntu-22.04.
+    - sh -c 'cd client/ui && wails3 generate bindings -clean=true -ts -f "-tags gtk3"'
+    - sh -c 'cd client/ui/frontend && pnpm install --frozen-lockfile && pnpm build'
+
+builds:
+  # Legacy GTK3 / WebKit2GTK 4.1 build for distros without WebKitGTK 6.0
+  # (Ubuntu 22.04, Debian 12, RHEL 9, Fedora <=39). The gtk3 tag flips the
+  # Wails Linux backend to the GTK3 stack and swaps our GTK4-only XEmbed
+  # tray host for the pure-Go stub (client/ui/xembed_host_gtk3_linux.go).
+  # Must be built on the oldest supported glibc (ubuntu-22.04 runner).
+  - id: netbird-ui-gtk3
+    dir: client/ui
+    binary: netbird-ui
+    env:
+      - CGO_ENABLED=1
+    goos:
+      - linux
+    goarch:
+      - amd64
+    ldflags:
+      - -s -w -X github.com/netbirdio/netbird/version.version={{.Version}} -X main.commit={{.Commit}} -X main.date={{.CommitDate}} -X main.builtBy=goreleaser
+    mod_timestamp: "{{ .CommitTimestamp }}"
+    tags:
+      - production
+      - gtk3
+
+archives:
+  - id: linux-gtk3-arch
+    name_template: "{{ .ProjectName }}-linux-gtk3_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
+    builds:
+      - netbird-ui-gtk3
+
+nfpms:
+  # Mutually-exclusive alternative to the GTK4 netbird-ui package -- both
+  # ship the same /usr/bin/netbird-ui from the shared stable/yum repos, so
+  # this one carries its own name and conflicts with the GTK4 package.
+  - maintainer: Netbird <dev@netbird.io>
+    description: Netbird client UI.
+    homepage: https://netbird.io/
+    license: BSD-3-Clause
+    vendor: NetBird
+    id: netbird_ui_deb_gtk3
+    package_name: netbird-ui-gtk3
+    file_name_template: "{{ .PackageName }}_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
+    builds:
+      - netbird-ui-gtk3
+    formats:
+      - deb
+    scripts:
+      postinstall: "release_files/ui-post-install.sh"
+    contents:
+      - src: client/ui/build/linux/netbird.desktop
+        dst: /usr/share/applications/org.wails.netbird.desktop
+      - src: client/ui/build/appicon.png
+        dst: /usr/share/pixmaps/netbird.png
+    conflicts:
+      - netbird-ui
+    replaces:
+      - netbird-ui
+    dependencies:
+      - netbird (>= 0.75.0)
+      - libgtk-3-0
+      - libwebkit2gtk-4.1-0
+      - xdg-utils
+
+  - maintainer: Netbird <dev@netbird.io>
+    description: Netbird client UI.
+    homepage: https://netbird.io/
+    license: BSD-3-Clause
+    vendor: NetBird
+    id: netbird_ui_rpm_gtk3
+    package_name: netbird-ui-gtk3
+    file_name_template: "{{ .PackageName }}_{{ .Version }}_{{ .Os }}_{{ .Arch }}"
+    builds:
+      - netbird-ui-gtk3
+    formats:
+      - rpm
+    scripts:
+      postinstall: "release_files/ui-post-install.sh"
+    contents:
+      - src: client/ui/build/linux/netbird.desktop
+        dst: /usr/share/applications/org.wails.netbird.desktop
+      - src: client/ui/build/appicon.png
+        dst: /usr/share/pixmaps/netbird.png
+    # No `replaces` here: nfpm maps it to rpm Obsoletes, which would make
+    # dnf swap installed GTK4 netbird-ui packages for this one on upgrade.
+    conflicts:
+      - netbird-ui
+    dependencies:
+      - netbird >= 0.75.0
+      - (gtk3 or libgtk-3-0)
+      - (webkit2gtk4.1 or libwebkit2gtk-4_1-0)
+      - xdg-utils
+
+    rpm:
+      signature:
+        key_file: '{{ if index .Env "GPG_RPM_KEY_FILE" }}{{ .Env.GPG_RPM_KEY_FILE }}{{ end }}'
+
+# The GTK4 UI job shares project_name, so the default checksum file name would
+# collide with it on the shared GitHub release.
+checksum:
+  name_template: "{{ .ProjectName }}_gtk3_checksums.txt"
+
+changelog:
+  disable: true
+
+uploads:
+  - name: debian
+    skip: "{{ .Env.SKIP_PUBLISH }}"
+    ids:
+      - netbird_ui_deb_gtk3
+    mode: archive
+    target: https://pkgs.wiretrustee.com/debian/pool/{{ .ArtifactName }};deb.distribution=stable;deb.component=main;deb.architecture={{ if .Arm }}armhf{{ else }}{{ .Arch }}{{ end }};deb.package=
+    username: dev@wiretrustee.com
+    method: PUT
+
+  - name: yum
+    skip: "{{ .Env.SKIP_PUBLISH }}"
+    ids:
+      - netbird_ui_rpm_gtk3
+    mode: archive
+    target: https://pkgs.wiretrustee.com/yum/{{ .Arch }}{{ if .Arm }}{{ .Arm }}{{ end }}
+    username: dev@wiretrustee.com
+    method: PUT
+
+release:
+  # Uploads into the release created by the main .goreleaser.yaml run.
+  # make_latest stays false everywhere: the signing pipeline
+  # (netbirdio/sign-pipelines) marks the release latest after the Windows
+  # and macOS artifacts are signed.
+  make_latest: false
+  prerelease: auto

AGENTS.md

@@ -0,0 +1,783 @@
+# NetBird Agent Guidelines
+
+**NetBird** is an open source connectivity platform: a WireGuard®-based overlay
+network with a control plane. The **agent** (`client/`) runs on user machines as
+a privileged daemon and manages the WireGuard interface, routing, firewall, and
+DNS. **Management** (`management/`) is the control plane and REST/gRPC API,
+**Signal** (`signal/`) brokers peer handshakes, **Relay** (`relay/`) carries
+traffic when a direct tunnel is impossible, and **Proxy** (`proxy/`) is the
+identity-aware proxy behind Agent Network.
+
+This file applies to the whole repository, and is the single source of truth for
+agent guidance here. `CLAUDE.md` is a one-line pointer to it — keep the guidance
+in this file, not duplicated there.
+
+## Contents
+
+- [STOP and ask the user before](#stop-and-ask-the-user-before)
+- [Quick reference](#quick-reference)
+- [Structure](#structure)
+- [Where to look](#where-to-look)
+- [Security](#security)
+- [Agent conventions](#agent-conventions)
+- [Repo-wide principles](#repo-wide-principles)
+- [Type safety](#type-safety)
+- [Concurrency and lifecycle](#concurrency-and-lifecycle)
+- [Error handling](#error-handling)
+- [Comments](#comments)
+- [Testing](#testing)
+- [Pitfalls](#pitfalls)
+- [Commits, PRs, releases](#commits-prs-releases)
+- [After you push: CI and review bots](#after-you-push-ci-and-review-bots)
+- [Discussion and support](#discussion-and-support)
+
+## STOP and ask the user before
+
+- **Opening a pull request for anything beyond a trivial fix, without an agreed
+  ticket.** Ask the user directly: *"Is there a discussion or issue for this
+  change?"* NetBird is discussion-first — community reports start in
+  [Discussions](https://github.com/netbirdio/netbird/discussions), DevRel
+  validates them, and only validated discussions become issues. A PR that
+  changes behavior with no linked issue may be closed on arrival. If there is no
+  ticket, offer to draft the discussion post **instead of** the PR, and wait for
+  the user's call. Only typos, broken links, documentation corrections, and
+  one-line fixes that already have an issue can skip this.
+- **Designing in any high-risk area** (see
+  [CONTRIBUTING.md](CONTRIBUTING.md#high-risk-areas)): public API and OpenAPI
+  schema, gRPC protos, behavior existing deployments would notice after an
+  upgrade, peer connectivity (ICE, NAT traversal, relay selection, WireGuard® or
+  Rosenpass key handling), client system integration (routing, firewall, DNS,
+  interface), authentication and authorization, CLI or service flags, config
+  file format, daemon IPC, store schema and migrations, or a new feature. The
+  design gets agreed in the ticket before code is written.
+- **Writing a store migration or changing a persisted model.** Migrations are
+  one-way in the field and both the GORM and pgx paths may need the change.
+- **Hand-editing generated code.** `*.pb.go`, `*.gen.go`, and mocks are outputs.
+  Edit the source (`.proto`, `openapi.yml`) and rerun the matching
+  `generate.sh`.
+- **Adding, removing, or bumping a dependency**, and never vendor a fork.
+- **Weakening a security control** — authentication, authorization, certificate
+  verification, privilege dropping, or peer identity checks — even when it is
+  the fastest way to make a test pass.
+- **Force-pushing to `main`**, force-pushing any branch that is already under
+  review, amending pushed commits, or bypassing hooks with `--no-verify`.
+
+## Quick reference
+
+```bash
+# Build
+go build ./...
+cd client && CGO_ENABLED=0 go build .        # agent
+cd management && go build .                  # management service
+cd signal && go build .                      # signal service
+
+# Verify (run before every push)
+go fmt ./...
+make lint            # golangci-lint on files changed vs origin/main (also the pre-push hook)
+make lint-all        # full-repository lint, matches CI
+make test-unit       # host-safe unit tests, -tags devcert, no sudo
+make test-privileged # privileged-tagged suite in a Docker container with NET_ADMIN
+make setup-hooks     # wire make lint into .githooks/pre-push
+
+# Narrow runs
+go test ./client/internal/dns/...
+go test -race -run TestPeerConn ./client/internal/peer/...
+PRIV_RUN=TestNftablesManager PRIV_PKGS=./client/firewall/nftables/... make test-privileged
+
+# Code generation (never hand-edit the output)
+./shared/management/http/api/generate.sh   # REST types from openapi.yml
+./shared/management/proto/generate.sh
+./shared/signal/proto/generate.sh
+./client/proto/generate.sh
+./flow/proto/generate.sh
+
+# Run locally (lab only, never on a machine you rely on)
+sudo ./client/netbird up --log-level debug --log-file console
+sudo ./client/netbird down                   # teardown: restores routing, firewall, DNS
+./signal/signal run --log-level debug --log-file console
+./management/management management --log-level debug --log-file console --config ./management.json
+```
+
+`netbird up` needs root and rewrites the host's routing table, firewall rules,
+DNS configuration, and WireGuard® interface. Run it only in a disposable test
+environment (a VM, container, or throwaway host) that you can rebuild, never on
+a workstation or server whose connectivity matters. Run `sudo netbird down`
+before you stop working, before rebuilding the binary, and on every failure
+path, so the host's networking state is restored instead of left half-applied.
+See [Pitfalls](#pitfalls) for why cleanup on every exit path matters.
+
+## Structure
+
+```text
+netbird/
+├── client/              NetBird agent
+│   ├── cmd/             agent CLI
+│   ├── internal/        agent business logic (engine, peer, dns, routemanager, ...)
+│   ├── server/          daemon for background execution
+│   ├── proto/           daemon gRPC protos
+│   ├── iface/           WireGuard® interface management
+│   ├── firewall/        nftables, iptables, pf, WFP, userspace backends
+│   ├── ssh/             built-in SSH server and client
+│   ├── ui/              desktop UI (Wails v3 + React)
+│   ├── android/, ios/   mobile bindings
+│   ├── wasm/            WebAssembly build
+│   └── mdm/, system/    MDM policy, host information
+├── management/          control plane
+│   └── server/          account, peer, groups, networks, posture, permissions,
+│                        settings, store, http (REST), idp, integrations, migration
+├── signal/              handshake broker (peer/, server/)
+├── relay/               relay service (protocol/, server/, healthcheck/)
+├── proxy/               identity-aware proxy (llm/, acme/, accesslog/, middleware/, tcp/, udp/)
+├── agent-network/       Agent Network overview
+├── shared/              imported by both agent and services
+│   ├── management/      proto/, client/, http/api (OpenAPI + generated types)
+│   ├── signal/          proto/, client/
+│   └── relay/, auth/, sshauth/, metrics/
+├── e2e/                 end-to-end suites and harness
+├── encryption/, dns/, route/, stun/, sharedsock/, util/, flow/
+├── infrastructure_files/  docker compose and getting-started templates
+└── release_files/       files packaged into releases
+```
+
+## Where to look
+
+| Task                        | Location                                                     |
+| --------------------------- | ------------------------------------------------------------ |
+| REST API / OpenAPI          | `shared/management/http/api/` + `management/server/http/`    |
+| Management gRPC protocol    | `shared/management/proto/`                                   |
+| Signal protocol             | `shared/signal/proto/`                                       |
+| Daemon IPC protocol         | `client/proto/`                                              |
+| Peer connection and NAT     | `client/internal/peer/`                                      |
+| Network map handling        | `client/internal/engine.go`, `shared/management/networkmap/` |
+| Routing                     | `client/internal/routemanager/`, `route/`                    |
+| Firewall backends           | `client/firewall/`                                           |
+| DNS                         | `client/internal/dns/`, `dns/`                               |
+| WireGuard® interface        | `client/iface/`                                              |
+| Persistence and migrations  | `management/server/store/`, `management/server/migration/`   |
+| IdP integrations            | `management/server/idp/`                                     |
+| Permissions model           | `management/server/permissions/`                             |
+| LLM routing / Agent Network | `proxy/internal/llm/`, `agent-network/`                      |
+| End-to-end tests            | `e2e/`                                                       |
+
+## Security
+
+### Never fail open
+
+When a security check — access control, an IP restriction, an auth decision —
+hits an error such as an unparseable value, an unavailable lookup, or a state it
+does not recognize, it must **deny**. Never skip the check or allow the request
+through because the check itself failed, and make the `default` and unknown cases
+of a security-related `switch` deny rather than fall through.
+
+### Daemon RPC input is untrusted
+
+The agent runs as root (LocalSystem on Windows), so a daemon RPC crosses a
+privilege boundary: treat every field as untrusted input rather than as something
+the UI or CLI validated on the way in.
+
+When you add or change an RPC, ask what the handler does with caller input while
+running as root. If the answer touches a filesystem path, a URL or host, or a
+privileged state change, it needs a gate **in the handler** — a check in the client
+that normally calls it is not a check at all.
+
+- **A caller-supplied path the daemon opens.** Never `os.Open` it as root.
+  Constrain it, then open it *as the caller* with `ipcauth.OpenOwnedFile`, which
+  opens `O_NOFOLLOW`, requires a regular file, and refuses a file the caller does
+  not own — so a symlink or hardlink aimed at a root-only file is rejected.
+- **A caller-supplied URL or host the daemon fetches.** Restrict the scheme and
+  allow only known hosts for unprivileged callers. Prefer a lexical host
+  allowlist plus TLS verification over "resolve the host, then reject private
+  IPs": the resolve-then-trust pattern has a DNS-rebinding race (public IP at
+  check time, attacker IP at connect time), while a name allowlist has no IP
+  check to race. Never accept `http://` where `https://` is expected.
+- **A privileged state change** (SSH root login, management URL, deregistration)
+  gates on the caller identity from `ipcauth.CallerIdentity(ctx)`.
+
+Caller identity comes from the kernel — `SO_PEERCRED`, `LOCAL_PEERCRED`, or the
+named-pipe client token — and never from an RPC field. When
+`ipcauth.CallerIdentity` reports that it could not determine an identity, **deny**;
+do not fall back to treating the caller as the transport peer.
+
+## Agent conventions
+
+### Three networking modes
+
+Where packets actually flow depends on the mode the agent is running in. The
+three are not interchangeable, so establish which one a change applies to — and
+what it should do in the other two — before you write it.
+
+- **kernel mode** (Linux only): in-kernel WireGuard®. The kernel handles both
+  peer-to-peer and routed traffic, and ACLs are iptables or nftables rules. The
+  client programs kernel facilities but never sees the traffic itself.
+- **userspace mode** (wireguard-go with a TUN): wireguard-go runs in-process. The
+  kernel handles peer-to-peer traffic once it leaves the TUN, while routed traffic
+  — exit nodes and network routes — goes through the userspace forwarder, which
+  terminates the connection and re-establishes it over OS sockets. Used on
+  platforms without kernel WireGuard® or when the user opts out.
+- **netstack mode**: wireguard-go in-process with no TUN and no kernel
+  networking. The forwarder does all routing by stitching userspace sockets, and
+  listeners such as the embedded SSH and DNS servers bind on a gVisor netstack.
+  Used where the process cannot create a TUN device, such as the embedded client
+  (`client/embed/`) and the WASM build.
+
+### The overlay interface is not "WireGuard"
+
+Do not put "WireGuard" in identifiers or comments unless the code is genuinely
+coupled to WireGuard® specifically — a wireguard-go call, a handshake field, a
+kernel WireGuard® netlink attribute. For the interface, the host, peers, or
+traffic in general, say "the NetBird interface", "the interface", or "the overlay".
+Most firewall, routing, and DNS code is transport-agnostic, so a WireGuard®
+reference there is simply inaccurate and rots as the transports change.
+
+### IPv6 is a soft feature
+
+The IPv6 overlay is opt-in dual-stack, and capability can change at runtime. Treat
+it as soft rather than a requirement:
+
+- Gate local v6 paths on the interface accessor (`wgIface.Address().HasIPv6()`),
+  not on raw state fields, and skip the v6 path when the host has no v6 rather
+  than returning an error.
+- Treat an empty or unparseable peer v6 address as "no v6 for that peer" and skip
+  it, keeping the v4 path working.
+- Never let a missing v6 break v4. Fail-closed is for security checks; a
+  capability mismatch skips the v6 work and carries on.
+
+### Environment variables
+
+Name the variable in a constant and parse booleans with `strconv.ParseBool` rather
+than comparing strings inline, so an unexpected value is logged instead of
+silently meaning false:
+
+```go
+const EnvDisableFeature = "NB_DISABLE_FEATURE"
+
+func isDisabledByEnv() bool {
+    val := os.Getenv(EnvDisableFeature)
+    if val == "" {
+        return false
+    }
+    disabled, err := strconv.ParseBool(val)
+    if err != nil {
+        log.Warnf("failed to parse %s: %v", EnvDisableFeature, err)
+        return false
+    }
+    return disabled
+}
+```
+
+### Validating against protocol specs
+
+When a change depends on what a protocol actually mandates, read the specification
+text from the [IETF datatracker](https://datatracker.ietf.org/) rather than a
+summary, and check that you have the current RFC — the widely cited one for a
+protocol is often superseded. Cite the section, not just the document, so a
+reviewer can jump straight to the rule.
+
+## Repo-wide principles
+
+1. **Run `go fmt` on every modified Go file.** Formatting is not optional.
+2. **Zero unaddressed linter warnings.** Fix what `golangci-lint` reports on code
+   you touch, and delete imports, helpers, and parameters your refactor orphaned.
+   Exception: unused parameters in shared code may be consumed by builds outside
+   this repository — do not remove them, ask instead.
+3. **Function comments are mandatory for exported functions**, written as full
+   sentences with a period, starting with the identifier name.
+4. **Prefer private functions and constants.** Export only what a caller outside
+   the package genuinely needs.
+5. **Early returns and guard clauses.** Handle errors and edge cases first
+   instead of nesting `if`/`else` chains.
+6. **Split complex functions.** If a function trips a complexity warning, break
+   it into named helpers rather than silencing the warning.
+7. **Avoid LLM-slop tells:** em dashes, hedging narration, restating the diff in
+   prose, trailing summaries. Defaults, not absolute bans. Applies to code,
+   comments, commit messages, and PR descriptions alike.
+8. **Concurrency: do a two-pass race analysis after every change** that touches
+   shared state, including reads of existing maps and slices. Guard them with a
+   mutex (or an atomic or channel where that fits better), keep critical
+   sections short, and run `go test -race` on the touched packages. See
+   [Concurrency and lifecycle](#concurrency-and-lifecycle) for the failure modes
+   to check for.
+9. **Cross-platform builds must keep working.** The agent targets Linux, macOS,
+   Windows, FreeBSD, Android, and iOS. When you add a platform-specific file,
+   add the counterpart or a build-tagged fallback for the others.
+10. **Never hand-edit generated files.** Change the source and regenerate.
+11. **Never log secrets** — private keys, setup keys, tokens, PAT values — and
+    keep peer IPs and hostnames out of logs above debug level.
+
+## Type safety
+
+**No bare primitives for domain concepts.** A `string` parameter for an account
+ID next to a `string` parameter for a peer ID is two bugs waiting to happen,
+because the compiler cannot catch the swap. Declare the type once and use it
+throughout, converting only at the boundaries where data enters or leaves —
+protobuf, gRPC, HTTP, an external library.
+
+```go
+type ServiceID string
+type AccountID string
+
+// Internal: typed all the way through
+func (r *Router) RemoveRoute(host SNIHost, svcID ServiceID) { ... }
+
+// Proto boundary: convert once, on the way in and on the way out
+svcID := ServiceID(mapping.GetId())
+req.ServiceId = string(svcID)
+```
+
+- **IP addresses are `netip.Addr`**, not `string` and not `net.IP`. Parse at the
+  boundary and pass the typed value inward.
+- **Always `Unmap()`** after parsing an address, after converting from `net.IP`,
+  and after extracting one from `RemoteAddr()`. This normalizes a v4-mapped v6
+  address (`::ffff:10.1.2.3`) to plain v4 so IPv4 rules match it. A stored or
+  compared mapped address silently fails to match those rules.
+- **Ports are `uint16`** internally; use `int` only where a library forces it and
+  convert immediately.
+- **Enums are a typed string with constants**, so the valid set is discoverable
+  and a typo fails to compile.
+- **Map keys follow the same rule**, and must be a real type (`type ServiceID
+  string`) rather than an alias (`type serviceID = string`) — an alias silently
+  accepts bare strings.
+
+## Concurrency and lifecycle
+
+Beyond the mutex hygiene in the principles above, check for these failure
+modes.
+
+- **Never read a struct field inside a goroutine** when another goroutine may nil
+  or reassign it. Pass the value as a parameter, or capture it into a local before
+  launching. This matters most when `Stop()` nils a field without waiting for the
+  goroutine to finish.
+
+  ```go
+  go func(ifaceName string) {   // good: passed in, cannot be nilled underneath
+      m.Start(ctx, ifaceName)
+  }(iface.Name())
+  ```
+
+- **Never wait on a channel while holding a lock the sender needs.** Copy what you
+  need out from under the lock, release it, then wait.
+
+  ```go
+  func (m *Manager) Stop() {
+      m.mu.Lock()
+      cancel, done := m.cancel, m.done
+      m.mu.Unlock()
+      if cancel != nil {
+          cancel()
+          <-done
+      }
+  }
+  ```
+
+- **`Stop`/`Close` must be idempotent** — guard on an already-stopped flag or a
+  nil cancel — and must release the state they guarded. Clear maps and caches;
+  a cancelled goroutine holding a live map still pins that memory. Note that a
+  nil map only panics on writes; reads and iteration behave like an empty map,
+  so where post-close use must be rejected, check the stopped flag explicitly.
+- **Publish coupled state only after every fallible step succeeds.** When several
+  fields form an invariant, build them into locals and assign them to the receiver
+  at the end. Assigning as you go leaves the object half-initialized when a later
+  step fails, so a readiness predicate reports ready while a coupled field is nil.
+  If an earlier step already had an external side effect — a created chain, an
+  opened handle, an inserted rule — roll it back before returning the error.
+- **Clean up what you own on constructor error paths.** Once a constructor has
+  started something, every later error path must undo it: cancel a goroutine and
+  wait for it to exit, stop a ticker, close a watcher. The object is never
+  returned, so its `Close` will never run.
+- **A failed `Start` must undo everything it started.** When a component brings up
+  several subsystems in sequence — connection manager, watchers, routing, DNS,
+  flow, persisted state — a failure partway through has to tear down the ones
+  already running, not just close the handle the error came from. Put the
+  already-started guard *before* that teardown path, so a rejected second `Start`
+  cannot dismantle the one that is running.
+
+## Error handling
+
+Use single-assignment form when the error is only needed inside the `if`:
+
+```go
+// Good
+if err := someCall(); err != nil {
+    return fmt.Errorf("context: %w", err)
+}
+
+// Bad - unnecessary split
+err := someCall()
+if err != nil {
+    return fmt.Errorf("context: %w", err)
+}
+```
+
+Use multiple assignment when the value is needed after the block:
+
+```go
+result, err := someCall()
+if err != nil {
+    return fmt.Errorf("context: %w", err)
+}
+```
+
+Add short, meaningful context, and **do not** start `fmt.Errorf` messages with
+obvious words like "failed to" or "error":
+
+```go
+// Good
+return fmt.Errorf("parse remote address: %w", err)
+return fmt.Errorf("listen on %s: %w", addr, err)
+
+// Bad
+return fmt.Errorf("failed to parse remote address: %w", err)
+return fmt.Errorf("error listening on %s: %w", addr, err)
+
+// "failed" is fine in log messages
+log.Debugf("failed to parse remote address: %v", err)
+```
+
+Skip the wrapping when a function only extracts or delegates and the wrap would
+add nothing:
+
+```go
+func parseAddr(addr string) (string, int, error) {
+    host, portStr, err := net.SplitHostPort(addr)
+    if err != nil {
+        return "", 0, err
+    }
+    // ...
+}
+```
+
+Log the errors you choose not to act on:
+
+- `log.Debugf()` for errors that do not affect program flow but help debugging.
+- `log.Tracef()` for very verbose errors that would otherwise spam logs.
+- **Never ignore** errors from writes, network sends, or critical cleanup.
+- Close errors may be ignored for read-only operations; log them at debug for
+  writes.
+
+**Do not log and return the same error.** It gets reported twice, from two places,
+and the second reader cannot tell whether it happened once or twice. Return it and
+let the caller decide. The exception is an API handler that has already written a
+response. Internal helpers return errors rather than logging and swallowing them.
+
+**Never return a typed nil as an error.** A nil `*MyError` stored in an `error`
+interface is not nil, so `err != nil` is true and callers take the failure path on
+success. Return the error only where it is actually set:
+
+```go
+if _, err := conn.Write(buf); err != nil {   // good
+    return err
+}
+return nil
+```
+
+**Accumulate with `multierror` when an operation should continue past individual
+failures** — teardown, cleanup, or setup where partial success is acceptable.
+`client/errors.FormatErrorOrNil` returns nil for an empty accumulator, so callers
+still see a plain nil on full success:
+
+```go
+func (m *Manager) Cleanup() error {
+    var merr *multierror.Error
+    for _, r := range m.resources {
+        if err := r.Close(); err != nil {
+            merr = multierror.Append(merr, fmt.Errorf("close %s: %w", r.Name, err))
+        }
+    }
+    return nberrors.FormatErrorOrNil(merr)
+}
+```
+
+| Scenario              | Approach              | Why                                       |
+| --------------------- | --------------------- | ----------------------------------------- |
+| Cleanup / teardown    | Accumulate            | Clean up as much as possible              |
+| Setup with rollback   | Abort on first error  | Partial state is invalid; undo what stuck |
+| Setup with partial OK | Accumulate            | Degraded operation is still useful        |
+
+## Comments
+
+Comment the **why**, never the **what**. Default to no comment, and add one only
+when a hidden constraint or workaround would surprise a future reader. Never
+reference the current task, PR, or your own changes in a comment.
+
+```go
+// Bad - trailing comments explaining the obvious
+defer localConn.Close() // Close the connection
+if err != nil {         // Check if error occurred
+
+// Good
+defer localConn.Close()
+
+// Good - explains a non-obvious constraint
+// Use incremental checksum update per RFC 1624 for performance.
+checksum = updateChecksum(checksum, oldPort, newPort)
+```
+
+### Length budget
+
+Neither of these is linter-enforced, so they are conventions the surrounding code
+mostly follows rather than hard limits:
+
+- **Around 90 characters per line.** Wrap the comment rather than running well past
+  it.
+- **Roughly 250 characters per comment**, about three wrapped lines. Doc comments
+  on exported identifiers may exceed it when the API genuinely needs the
+  explanation; inline comments inside a function body rarely should.
+
+The budget is a smell detector, not a rule to game. Do not compress a needed
+explanation into cryptic shorthand to fit — if a block of code needs more than
+250 characters of prose, the code is doing too much. Fix the code:
+
+- **Extract a named function.** A well-named function replaces the comment: the
+  name says *what*, the body shows *how*, and the comment you no longer write
+  was the *what* anyway. Clean Code calls this "explain yourself in code".
+- **Extract a named constant or predicate.** `if isExpiredSetupKey(key)` needs
+  no comment; `if key.ExpiresAt.Before(now) && !key.Revoked && key.UsageLimit > 0`
+  does.
+- **Keep the surviving comment for the why** — the RFC, the kernel quirk, the
+  ordering constraint. That part is usually one or two lines.
+
+### Long switch and if/else chains
+
+A `switch` whose cases carry multi-line explanations is the usual place this
+budget is breached, and the comment is a symptom. In order of preference:
+
+1. **Extract each case body into a named function.** The case becomes one line,
+   the name carries the meaning, and the switch reads as a table of contents.
+2. **Replace the switch with a lookup table** — `map[Kind]handlerFunc` — when the
+   branches are uniform. Adding a case stops meaning editing a growing function.
+3. **Replace conditional with polymorphism** when branches vary by type and the
+   same switch shape starts appearing in more than one place. Clean Code's rule
+   of thumb: tolerate a switch statement if it appears **once**, is buried in a
+   factory that returns an interface, and no other switch dispatches on the same
+   type. A second switch over the same enum is the signal to introduce the
+   interface.
+
+Do not restructure a switch purely to satisfy the budget when the cases are one
+line each and self-evident — a flat, boring `switch` over an enum is fine and
+needs no comments at all.
+
+Explanatory comments in tests are welcome — they document the scenario being set
+up, and the 250-character budget does not apply to them.
+
+## Testing
+
+- **Unit tests** live beside the code as `_test.go`. `make test-unit` runs the
+  host-safe set with `-tags devcert` and no sudo.
+- **Privileged tests** carry the `privileged` build tag and mutate host
+  networking. They run through `make test-privileged`, inside a Docker container
+  with `NET_ADMIN`. Never bypass that harness by running them directly on the
+  host.
+- **End-to-end suites** live in `e2e/` with a shared harness.
+- **Test real behavior, not API existence.** Assert on the observable end state
+  a consumer would see — bytes that arrived, the packet after translation, the
+  row after the write — not merely that a method exists or returns an error.
+- **Avoid mocks for code we own.** Exercise the real store, manager, or
+  controller and assert what the caller actually receives.
+- **`require` for setup and preconditions, `assert` for the conditions under
+  test.** Use `require` whenever a later line would panic or be meaningless
+  otherwise.
+- **Message guidance:** optional for `NoError`/`Error`; always give context for
+  comparison, boolean, and collection assertions.
+- **Reproduce a bug before fixing it.** Write the test, watch it fail *for the
+  reason you expect* — a test that fails for an unrelated reason proves nothing —
+  then apply the fix and confirm it passes. Add the thin surrounding cases while
+  you are there.
+- **Use `t.Setenv`** rather than `os.Setenv` so the previous value is restored on
+  cleanup. To test the unset case, call `t.Setenv` first to register the restore,
+  then `os.Unsetenv`.
+- **Prefer `t.Cleanup` over `defer`** in any test with parallel subtests: the
+  parent function returns, running its `defer`s, while parallel subtests are
+  still suspended. Sequential subtests finish inside `t.Run`, so `defer` is safe
+  there, but `t.Cleanup` works in both cases.
+- **Explanatory comments in tests are welcome.** Describe the scenario being set
+  up; the comment budget below does not apply to them.
+
+```go
+server, err := StartTestServer()
+require.NoError(t, err, "Test server setup must succeed")
+defer server.Close()
+
+result, err := client.DoOperation()
+assert.NoError(t, err)
+assert.Equal(t, expectedResult, result, "Result should match expected")
+```
+
+## Pitfalls
+
+- **The agent runs as root.** Anything touching routing, firewall, DNS, or the
+  interface can take a user's machine off the network. Prefer a reversible
+  change and make sure cleanup runs on every exit path.
+- **Management has two account loaders** (GORM and pgx). Adding a relation to an
+  account often means updating both, or it silently comes back empty in
+  production.
+- **`go test ./...` without `-tags devcert` skips tests** that need the
+  development certificate. Use `make test-unit`.
+- **`make lint` only checks the diff against `origin/main`.** CI runs
+  `make lint-all`; run it too before pushing a large change.
+- **Protos are consumed by released clients.** An old agent must keep working
+  against a new Management, so fields are added, never renumbered or removed.
+- **Windows requires the wintun driver**, and the daemon serves a named pipe
+  (`npipe://netbird`) rather than loopback TCP. Loopback TCP carries no caller
+  identity, so privileged operations are refused over it.
+
+## Commits, PRs, releases
+
+- **PR titles must start with a bracketed tag.** Before you propose a title,
+  **read [`.github/workflows/pr-title-check.yml`](.github/workflows/pr-title-check.yml)
+  and take the allowed tags from the `allowedTags` array in that file.** It is
+  the only source of truth, it changes as components are added, and the check
+  runs on every title edit — a tag that is not in that array is a red build. Do
+  not rely on a list memorized from anywhere else, including this file.
+
+  ```text
+  [client] Authorize daemon IPC callers by their local identity
+  [management,client] Add MDM policy support
+  ```
+
+  Multiple tags are comma-separated inside one pair of brackets. Match the tag
+  to the component you actually changed, not to the one you read the most.
+
+- **Use the repository's PR template.** Fill in
+  [`.github/pull_request_template.md`](.github/pull_request_template.md) rather
+  than replacing it with your own summary: describe the change, link the issue,
+  tick the checklist honestly (including "ran locally" and "single purpose"),
+  and complete the documentation section. Do not tick a box you have not
+  verified, and do not delete rows that do not apply — the docs gate in CI reads
+  that section and fails when it is missing.
+
+- **Keep the PR description short.** Under 1000 words on top of the template's
+  own text, and usually far less — a few paragraphs. Reviewers read the diff;
+  the description exists to explain what the diff cannot say for itself. This is
+  well below what an agent will produce by default, so cut before you post.
+
+- **Body: why before what.** Lead with the problem and the reason for this
+  approach, then the shape of the change. No bullet list of files changed, no
+  per-function walkthrough, no restating the diff in prose, no trailing summary
+  section, no self-congratulatory closing line.
+
+- **No `Co-Authored-By` or tool-attribution trailers in the PR description**,
+  and none in commits either. Contributors own their contributions. Whatever
+  tooling produced the diff, the person opening the PR is its author: they have
+  read every line, they can explain why it works, they can answer review
+  questions without going back to a model, and they are accountable for the
+  consequences of merging it. Do not add a trailer, footer, or description line
+  that spreads that ownership onto a tool.
+
+- **Commit subjects follow the same `[scope] Subject` convention.** Keep the
+  subject short, and use the body for why before what. No bullet lists of files
+  changed.
+
+- **Push review fixes as separate commits.** The PR is squashed on merge, so
+  there is no reason to rewrite history mid-review; many small commits make the
+  re-review readable.
+
+- **Do not force-push a branch that is under review.** A force-push detaches
+  existing review comments from the lines they were written against, destroys
+  the "changes since your last review" diff a reviewer relies on, and discards
+  the CI history that showed which commit broke what. Add commits instead —
+  including for fixups and reverts. Force-push only when there is no
+  alternative: a rebase to clear a genuine conflict, or removing a secret or a
+  large binary that was committed by mistake. When you must, ask the user first,
+  then say so in a PR comment so reviewers know their anchors moved. Never
+  force-push `main`, and never force-push a branch you do not own.
+
+- **One PR, one purpose.** Split refactors out of fixes and fixes out of
+  features.
+
+- **Keep the PR small.** Size is the single strongest predictor of how long a PR
+  waits. Aim for **under ~400 changed lines across under ~20 files**; past
+  roughly **1000 lines or 50 files** a community PR is likely to be sent back to
+  be split, or left unreviewed until it is. Large PRs from outside the core team
+  may be blocked outright when the size was never agreed in the ticket —
+  reviewing a sprawling change against a privileged networking daemon is a
+  security risk in itself, not just a time cost.
+
+  Judge the size by hand-written code: exclude generated output, `go.sum`,
+  vendored files, and test fixtures from the estimate, but do not use their
+  presence to argue a 3000-line PR is small.
+
+  When a change genuinely cannot be small — a protocol migration, a
+  cross-component rename — agree the split in the ticket **before** writing
+  code, and land it as a sequence of PRs that each build, test, and make sense
+  on their own. Propose that split to the user rather than opening one large PR
+  and hoping.
+
+  Prefer GitHub's stacked pull requests for such a sequence, rather than
+  hand-managing base branches: open each PR against the branch below it instead of
+  `main`, so every PR's diff shows only its own change. Merging a layer retargets
+  the PRs above it, and branch protections and required checks on the base branch
+  still apply to each one.
+
+- **User-facing changes need a docs PR** in
+  [netbirdio/docs](https://github.com/netbirdio/docs), linked from the PR
+  description.
+
+## After you push: CI and review bots
+
+Opening the PR is not the end of the task. Watch the run, read what the bots
+say, and drive the PR to green before you report the work as done.
+
+```bash
+gh pr checks <pr>            --watch    # all checks, live
+gh run view <run-id> --log-failed       # only the failing steps
+gh pr view <pr> --comments              # bot and human review comments
+```
+
+**Never report a change as finished while checks are pending or red**, and never
+describe a red PR as passing. If you ran out of turn before CI finished, say
+which checks were still running.
+
+### The checks
+
+- **Go tests** — `golang-test-{linux,darwin,windows,freebsd}.yml`, sharded per
+  component. A failure in a component you did not touch is usually a real
+  interaction, not noise; read the log before assuming flake.
+- **golangci-lint** — `golangci-lint.yml` runs the full repository, while
+  `make lint` only checks your diff. A clean local lint does not guarantee green
+  CI on a large change.
+- **PR Title Check** — `pr-title-check.yml`, see above.
+- **Codecov** — uploaded from the Linux test workflow with per-component flags
+  (`unit,client`, `unit,management`, `unit,relay`, `unit,proxy`, `unit,signal`,
+  `integration,management`). Coverage on new code should not go backwards. Add
+  tests for the paths you introduced; do not adjust thresholds or exclude files
+  to clear the report.
+- **CodeRabbit** — configured in [`.coderabbit.yaml`](.coderabbit.yaml): `chill`
+  profile, auto-review on every non-draft PR, TypeScript/JavaScript/SVG paths
+  filtered out. Chat auto-reply is on, so `@coderabbitai` in a comment reaches
+  it.
+- **SonarCloud** — project `netbirdio_netbird`, quality gate on new code (bugs,
+  vulnerabilities, code smells, duplication, coverage).
+- **Snyk** — dependency and code scanning.
+
+Sonar and Snyk report as GitHub App checks rather than workflows in this
+repository, so their detail lives on the PR check, not in the Actions logs.
+
+### Handling bot findings
+
+- **Read every comment and act on it.** Either fix it, or reply with the reason
+  it does not apply. Do not bulk-resolve threads to clear the count, and do not
+  silently ignore a finding because the check is advisory.
+- **Bots are frequently wrong here.** NetBird has privileged, platform-specific,
+  and concurrency-heavy code that static analysis reads poorly. A confident
+  CodeRabbit or Sonar comment can still be nonsense. Verify the claim against
+  the code before you change anything — never edit correct code just to silence
+  a bot.
+- **Security findings get the opposite default.** For a Snyk or Sonar
+  vulnerability, or a CodeRabbit comment about authentication, authorization,
+  certificate verification, or key handling, assume it is real until you have
+  disproved it. Surface it to the user rather than dismissing it yourself.
+- **A new vulnerable dependency is a stop.** Bumping or replacing dependencies
+  needs the user's decision, as above.
+- **Never change a workflow, threshold, lint exclusion, or bot config to make a
+  check pass.** If a check is genuinely wrong, say so and let the user decide.
+- **Do not paper over flakes with blind re-runs.** Identify the failure first. If
+  it is a known flake, name it; if you cannot tell, report it as unresolved
+  rather than re-running until it goes green.
+
+## Discussion and support
+
+- Discussions: <https://github.com/netbirdio/netbird/discussions>
+- Slack: <https://docs.netbird.io/slack-url>
+- Docs: <https://docs.netbird.io>
+- Security: <https://github.com/netbirdio/netbird/security/policy> — never in public
+- Contribution process: [CONTRIBUTING.md](CONTRIBUTING.md)

CLAUDE.md

@@ -0,0 +1 @@
+See [AGENTS.md](AGENTS.md) for the agent guidelines in this repository.

CONTRIBUTING.md

@@ -66,11 +66,41 @@ Typical bug fixes, internal refactors, documentation updates, and tests do not
 need a design discussion, but should still be tied to an issue so the work is
 visible and nobody duplicates it.
 
+### Using AI coding agents
+
+We have no policy for or against using an AI agent to write NetBird code. That
+choice is yours, and we are not going to interrogate anyone about their tools.
+
+What we do have is a lot of incoming contributions that were plainly drafted with
+one, and enough experience reviewing them to see the same avoidable problems
+again and again: no ticket behind the change, a diff far too large to review, a
+description longer than the code it describes, an approach that was never going
+to be accepted, and an author who cannot answer questions about their own PR.
+None of that is caused by the tooling — it is what happens when a tool is pointed
+at a repository whose expectations it has never been told.
+
+So rather than a rule, there is a guide. [AGENTS.md](AGENTS.md) restates the
+expectations from this document in the form agents read automatically
+(`CLAUDE.md` points to it), so pointing your tool at the repository is usually
+enough. Among other things it tells the agent to ask you for the
+discussion or issue before drafting a PR, to keep the change small and
+single-purpose, to run the tests locally, to use this repository's PR template
+and title tags, and to write a description a reviewer can get through.
+
+The guardrails are the point, and they are the same ones we apply to everyone: an
+agreed ticket, a change you have actually run, a diff small enough to review with
+care, and an author who can explain it. Whatever wrote the diff, you are its
+author — you own every line you submit and the consequences of opening a PR with it.
+
+We may assess whether a contribution is maintainable and whether its merged code
+aligns with our security standards and design expectations.
+
 ## Contents
 
 - [Contributing to NetBird](#contributing-to-netbird)
     - [Ticket first, PR second](#ticket-first-pr-second)
         - [High-risk areas](#high-risk-areas)
+        - [Using AI coding agents](#using-ai-coding-agents)
     - [Contents](#contents)
     - [Code of conduct](#code-of-conduct)
     - [Directory structure](#directory-structure)
@@ -82,6 +112,7 @@ visible and nobody duplicates it.
         - [Test suite](#test-suite)
     - [Checklist before submitting a PR](#checklist-before-submitting-a-pr)
     - [When we close a PR](#when-we-close-a-pr)
+    - [Translations](#translations)
     - [Other project repositories](#other-project-repositories)
     - [Contributor License Agreement](#contributor-license-agreement)
 
@@ -161,7 +192,7 @@ dependencies are installed. Here is a short guide on how that can be done.
 
 ### Requirements
 
-#### Go 1.25
+#### Go 1.26
 
 Follow the installation guide from https://go.dev/
 
@@ -169,7 +200,7 @@ Follow the installation guide from https://go.dev/
 
 The desktop UI client (`client/ui`) is built with [Wails v3](https://v3.wails.io/) and a React frontend rendered in a WebView. To build it you need:
 
-- Go ≥ 1.25
+- Go ≥ 1.26
 - Node ≥ 20 and **pnpm** (`corepack enable && corepack prepare pnpm@latest --activate`)
 - The `wails3` CLI: `go install github.com/wailsapp/wails/v3/cmd/wails3@latest`
 - The `task` runner: `go install github.com/go-task/task/v3/cmd/task@latest`
@@ -448,7 +479,7 @@ go test -race ./client/internal/dns/...
 
 ## Checklist before submitting a PR
 
-As a critical network service and open-source project, we must enforce a few
+As a critical network service and open source project, we must enforce a few
 things before submitting a pull request. The
 [pull request template](/.github/pull_request_template.md) mirrors this list —
 fill it in rather than deleting it.
@@ -582,6 +613,17 @@ A closed PR is not a rejected idea. Take it back to the
 [discussion](https://github.com/netbirdio/netbird/discussions), settle the
 approach, and reopen the work from there.
 
+## Translations
+
+Desktop UI translations are not contributed through pull requests. Translate on
+[Crowdin](https://crowdin.com/project/netbird) instead: no ticket needed, just
+join the project and pick your language. Crowdin syncs with this repository and
+opens the service PRs itself, so hand-edited locale files would conflict with
+the next sync. Style, terminology, and review guidance live in
+[client/ui/i18n/TRANSLATING.md](client/ui/i18n/TRANSLATING.md). To request a
+language the project does not offer yet, ask on the Crowdin project page or in
+a [discussion](https://github.com/netbirdio/netbird/discussions).
+
 ## Other project repositories
 
 NetBird project is composed of 3 main repositories:

README.md

@@ -26,7 +26,7 @@
   <strong>
     Start using NetBird at <a href="https://netbird.io/pricing">netbird.io</a>
     <br/>
-    See <a href="https://netbird.io/docs/">Documentation</a>
+    See <a href="https://docs.netbird.io/">Documentation</a>
     <br/>
     Join our <a href="https://docs.netbird.io/slack-url">Slack channel</a> or our <a href="https://forum.netbird.io">Community forum</a>
   </strong>
@@ -130,7 +130,7 @@ In November 2022, NetBird joined the [StartUpSecure program](https://www.forschu
 ![CISPA_Logo_BLACK_EN_RZ_RGB (1)](https://user-images.githubusercontent.com/700848/203091324-c6d311a0-22b5-4b05-a288-91cbc6cdcc46.png)
 
 ### Acknowledgements
-We build on open-source technologies like [WireGuard®](https://www.wireguard.com/), [Pion ICE](https://github.com/pion/ice), and [Rosenpass](https://rosenpass.eu). We greatly appreciate the work these projects are doing, and we'd love it if you could support them too (e.g., by starring or contributing).
+We build on open source technologies like [WireGuard®](https://www.wireguard.com/), [Pion ICE](https://github.com/pion/ice), and [Rosenpass](https://rosenpass.eu). We greatly appreciate the work these projects are doing, and we'd love it if you could support them too (e.g., by starring or contributing).
 
 ### Legal
 This repository is licensed under the BSD-3-Clause license, which applies to all parts of the repository except for the directories management/, signal/ and relay/.

SECURITY.md

@@ -14,7 +14,7 @@ Report security issues one of these two ways:
   on this repository. This is the preferred route: it keeps the discussion, the draft advisory, and the credit in one place.
 - **Email** — `security@netbird.io`.
 
-If the finding affects NetBird Cloud or our hosted infrastructure rather than the open-source code, email us rather than
+If the finding affects NetBird Cloud or our hosted infrastructure rather than the open source code, email us rather than
 filing a repository report.
 
 ### What to include

agent-network/README.md

@@ -40,6 +40,35 @@ You can then use this private endpoint to configure your AI agents, whether that
 Full step-by-step setup:
 **https://docs.netbird.io/agent-network/quickstart**
 
+## Client settings that don't follow the endpoint
+
+Most of an agent's traffic follows the base URL you hand it, but a few
+client-side checks call their vendor directly and never reach the proxy. On a
+network that blocks direct egress they fail even though inference works, so
+they are worth setting once when you roll the endpoint out.
+
+For Claude Code:
+
+- **Fast mode** checks availability against `api.anthropic.com` rather than the
+  configured base URL. Set `CLAUDE_CODE_SKIP_FAST_MODE_ORG_CHECK=1` when the
+  agent authenticates with `ANTHROPIC_AUTH_TOKEN` alone (the usual shape when
+  the proxy injects the real provider key) or when a TLS-inspecting proxy
+  answers the check itself. Set
+  `CLAUDE_CODE_SKIP_FAST_MODE_NETWORK_ERRORS=1` when the network refuses the
+  connection outright. Fast mode is an Anthropic-API feature, so it is
+  unavailable on a Bedrock- or Vertex-backed endpoint whatever you set.
+- **Model discovery** is off by default. Set
+  `CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1` for the picker to list the
+  models your policies authorise; the proxy filters the response to that set.
+  The client gives discovery a three-second budget and treats any redirect as
+  a failure, so the endpoint must serve `/v1/models` directly.
+- **The WebFetch domain safety check** also calls `api.anthropic.com` directly
+  and is unaffected by the variables above.
+
+Allowing direct egress to `api.anthropic.com` covers the network cases but not
+the credential one, where the check reaches Anthropic and is rejected because
+the agent presents a proxy-issued key.
+
 ## Architecture 
 
 Agent Network is built on two existing NetBird capabilities:
@@ -67,6 +96,42 @@ components:
   — the management-side control plane: providers, policies, guardrails, limits, routing,
   and usage/access logs.
 
+## Access roles
+
+Agent Network permissions build on the account permission matrix
+([`management/server/permissions/`](../management/server/permissions)). The
+`agent_network` area is split into dotted submodules (`agent_network.providers`,
+`.policies`, `.guardrails`, `.budgets`, `.usage`, `.logs`, `.settings`); a role may
+grant a single submodule or the parent, which cascades to all of them.
+
+Two roles delegate Agent Network access without account-admin rights:
+
+- **`agent_network_admin`** — full control over the whole `agent_network` area plus
+  read-only users, groups, peers, and account info (needed to build policies).
+  Nothing else in the account.
+- **`usage_viewer`** — the regular User baseline plus read on
+  `agent_network.usage` (the aggregated usage and cost overview) and read-only
+  access to the resources the usage filters resolve against: users, groups,
+  peers, and the provider list (connection config redacted — no upstream URLs
+  or operator-supplied header values). No policies, and no account-wide
+  request-level access logs; like any caller, it still reads its own requests
+  through the self-scoped endpoints below.
+
+Every authenticated user, regardless of role, can read the caller-scoped
+self-service endpoint `GET /api/agent-network/agent-config` (the endpoint, providers,
+and models the caller's own policies allow — what a local AI tool needs and nothing
+more). The regular usage and access-log endpoints self-scope instead of denying:
+a caller without the account-wide grant gets their own rows back, so "my usage"
+and "my requests" are the same endpoints the admin dashboard uses. The provider
+list self-scopes the same way — a caller without the providers grant gets the
+providers their own policies authorize, reduced to the display surface, with
+each provider's model list cut to what the caller's policy guardrails and the
+provider's declared models effectively permit (the same computation the setup
+answer and the proxy use). This feeds the dashboard's provider and model
+filters. Role
+definitions live in
+[`management/server/permissions/roles/`](../management/server/permissions/roles).
+
 ## Documentation
 
 Full documentation, architecture, and quickstart:

client/android/client.go

@@ -15,6 +15,7 @@ import (
 
 	log "github.com/sirupsen/logrus"
 
+	nbAnonymize "github.com/netbirdio/netbird/client/anonymize"
 	"github.com/netbirdio/netbird/client/iface/device"
 	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/debug"
@@ -25,6 +26,7 @@ import (
 	"github.com/netbirdio/netbird/client/internal/routemanager"
 	"github.com/netbirdio/netbird/client/internal/stdnet"
 	"github.com/netbirdio/netbird/client/net"
+	"github.com/netbirdio/netbird/client/netevents"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/formatter"
 	"github.com/netbirdio/netbird/route"
@@ -32,10 +34,12 @@ import (
 	types "github.com/netbirdio/netbird/upload-server/types"
 )
 
-// ConnectionListener export internal Listener for mobile
-type ConnectionListener interface {
-	peer.Listener
-}
+// AnonymizeLevelDefault and AnonymizeLevelStrict are the accepted
+// anonymizeLevel values for DebugBundle.
+const (
+	AnonymizeLevelDefault = nbAnonymize.LevelDefaultString
+	AnonymizeLevelStrict  = nbAnonymize.LevelStrictString
+)
 
 // TunAdapter export internal TunAdapter for mobile
 type TunAdapter interface {
@@ -57,6 +61,12 @@ type DnsReadyListener interface {
 	dns.ReadyListener
 }
 
+// TunSettings is a snapshot of the settings the TUN device is rebuilt with
+type TunSettings struct {
+	Routes        string
+	SearchDomains string
+}
+
 func init() {
 	formatter.SetLogcatFormatter(log.StandardLogger())
 }
@@ -71,11 +81,17 @@ type Client struct {
 	deviceName            string
 	uiVersion             string
 	networkChangeListener listener.NetworkChangeListener
+	// netMgr outlives engine restarts: it mirrors the OS connectivity, not
+	// the engine lifecycle. Run and RunWithoutLogin inject its state and
+	// sweeper into each new ConnectClient.
+	netMgr *netevents.Manager
 
 	stateMu       sync.RWMutex
 	connectClient *internal.ConnectClient
 	config        *profilemanager.Config
 	cacheDir      string
+	// Identifies the running profile for the SSO login hint; see profile_state.go.
+	cfgPath string
 
 	stateChangeMu    sync.Mutex
 	stateChangeSubID string
@@ -96,11 +112,12 @@ type Client struct {
 	extendCancel context.CancelFunc
 }
 
-func (c *Client) setState(cfg *profilemanager.Config, cacheDir string, cc *internal.ConnectClient) {
+func (c *Client) setState(cfg *profilemanager.Config, cacheDir string, cfgPath string, cc *internal.ConnectClient) {
 	c.stateMu.Lock()
 	defer c.stateMu.Unlock()
 	c.config = cfg
 	c.cacheDir = cacheDir
+	c.cfgPath = cfgPath
 	c.connectClient = cc
 }
 
@@ -110,6 +127,16 @@ func (c *Client) stateSnapshot() (*profilemanager.Config, string, *internal.Conn
 	return c.config, c.cacheDir, c.connectClient
 }
 
+// authSnapshot returns the config together with the path it was loaded from, in
+// one lock: the path identifies the profile whose account email backs the login
+// hint, so reading it separately could pair one profile's config with another's
+// hint when a profile switch lands in between.
+func (c *Client) authSnapshot() (*profilemanager.Config, string, *internal.ConnectClient) {
+	c.stateMu.RLock()
+	defer c.stateMu.RUnlock()
+	return c.config, c.cfgPath, c.connectClient
+}
+
 func (c *Client) getConnectClient() *internal.ConnectClient {
 	c.stateMu.RLock()
 	defer c.stateMu.RUnlock()
@@ -121,14 +148,17 @@ func NewClient(androidSDKVersion int, deviceName string, uiVersion string, tunAd
 	execWorkaround(androidSDKVersion)
 
 	net.SetAndroidProtectSocketFn(tunAdapter.ProtectSocket)
+	system.SetIFaceDiscover(iFaceDiscover)
+	recorder := peer.NewRecorder("")
 	return &Client{
 		deviceName:            deviceName,
 		uiVersion:             uiVersion,
 		tunAdapter:            tunAdapter,
 		iFaceDiscover:         iFaceDiscover,
-		recorder:              peer.NewRecorder(""),
+		recorder:              recorder,
 		ctxCancelLock:         &sync.Mutex{},
 		networkChangeListener: networkChangeListener,
+		netMgr:                netevents.NewManager(recorder),
 	}
 }
 
@@ -162,15 +192,17 @@ func (c *Client) Run(platformFiles PlatformFiles, urlOpener URLOpener, isAndroid
 	defer c.ctxCancel()
 	c.ctxCancelLock.Unlock()
 
-	auth := NewAuthWithConfig(ctx, cfg)
+	auth := NewAuthWithConfig(ctx, cfg, cfgFile)
 	err = auth.login(urlOpener, isAndroidTV)
 	if err != nil {
 		return err
 	}
 	// todo do not throw error in case of cancelled context
 	ctx = internal.CtxInitState(ctx)
-	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder)
-	c.setState(cfg, cacheDir, connectClient)
+
+	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder,
+		internal.WithNetEvents(c.netMgr))
+	c.setState(cfg, cacheDir, cfgFile, connectClient)
 	// This path runs the interactive SSO flow, so reaching here means the peer
 	// is authenticated again — release the latch Status() reports from. Clear
 	// only once the fresh connect client is installed: until then Status()
@@ -210,8 +242,9 @@ func (c *Client) RunWithoutLogin(platformFiles PlatformFiles, dns *DNSList, dnsR
 
 	// todo do not throw error in case of cancelled context
 	ctx = internal.CtxInitState(ctx)
-	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder)
-	c.setState(cfg, cacheDir, connectClient)
+	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder,
+		internal.WithNetEvents(c.netMgr))
+	c.setState(cfg, cacheDir, cfgFile, connectClient)
 	return connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile, cacheDir)
 }
 
@@ -240,9 +273,49 @@ func (c *Client) RenewTun(fd int) error {
 	return e.RenewTun(fd)
 }
 
+func (c *Client) GetTunSettings() (*TunSettings, error) {
+	cc := c.getConnectClient()
+	if cc == nil {
+		return nil, fmt.Errorf("engine not running")
+	}
+
+	e := cc.Engine()
+	if e == nil {
+		return nil, fmt.Errorf("engine not initialized")
+	}
+
+	routes, searchDomains := e.TunSettings()
+	return &TunSettings{
+		Routes:        strings.Join(routes, ";"),
+		SearchDomains: strings.Join(searchDomains, ";"),
+	}, nil
+}
+
+// SetNetworkAvailable feeds OS-reported network availability into the client.
+// While unavailable, the internal reconnect loops suspend their attempts and
+// the connection listener reports NoNetwork instead of Connecting; when
+// availability returns, the loops resume immediately with a fresh backoff.
+// Losing the last network also sweeps the registered connections: nothing can
+// redial while offline, so the stale sockets would otherwise stay silently
+// "connected" until their own timeouts and the client would keep reporting
+// Connected with no network at all.
+func (c *Client) SetNetworkAvailable(available bool) {
+	c.netMgr.SetNetworkAvailable(available)
+}
+
+// NotifyNetworkChange marks the management, signal and relay connections
+// stale after the OS switched networks and schedules a sweep that cuts
+// whatever has not redialed on the new network by then. The engine and the
+// TUN device stay untouched.
+func (c *Client) NotifyNetworkChange() {
+	c.netMgr.NotifyNetworkChange()
+}
+
 // DebugBundle generates a debug bundle, uploads it, and returns the upload key.
-// It works both with and without a running engine.
-func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (string, error) {
+// It works both with and without a running engine. anonymizeLevel is "default"
+// or "strict"; strict also anonymizes internal IP ranges, peer names, and
+// WireGuard public keys, and implies anonymize.
+func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool, anonymizeLevel string) (string, error) {
 	cfg, cacheDir, cc := c.stateSnapshot()
 
 	// If the engine hasn't been started, load config from disk
@@ -261,6 +334,7 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
 		InternalConfig: cfg,
 		StatusRecorder: c.recorder,
 		TempDir:        cacheDir,
+		StatePath:      platformFiles.StateFilePath(),
 	}
 
 	if cc != nil {
@@ -284,6 +358,7 @@ func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (strin
 		deps,
 		debug.BundleConfig{
 			Anonymize:         anonymize,
+			AnonymizeLevel:    nbAnonymize.ParseLevel(anonymizeLevel),
 			IncludeSystemInfo: true,
 		},
 	)
@@ -476,7 +551,11 @@ func (c *Client) OnUpdatedHostDNS(list *DNSList) error {
 
 // SetConnectionListener set the network connection listener
 func (c *Client) SetConnectionListener(listener ConnectionListener) {
-	c.recorder.SetConnectionListener(listener)
+	if listener == nil {
+		c.recorder.RemoveConnectionListener()
+		return
+	}
+	c.recorder.SetConnectionListener(connectionListenerAdapter{listener})
 }
 
 // RemoveConnectionListener remove connection listener

client/android/connection_listener.go

@@ -0,0 +1,41 @@
+//go:build android
+
+package android
+
+import (
+	"github.com/netbirdio/netbird/client/internal/peer"
+)
+
+// Client state values delivered via ConnectionListener.OnStateChanged,
+// re-exported as basic constants so gomobile emits them into the generated
+// Java bindings. They mirror peer.ClientState*: append-only, never reorder.
+const (
+	ClientStateDisconnected  = int(peer.ClientStateDisconnected)
+	ClientStateConnected     = int(peer.ClientStateConnected)
+	ClientStateConnecting    = int(peer.ClientStateConnecting)
+	ClientStateDisconnecting = int(peer.ClientStateDisconnecting)
+	ClientStateNoNetwork     = int(peer.ClientStateNoNetwork)
+)
+
+// ConnectionListener export internal Listener for mobile. It mirrors
+// peer.Listener with OnStateChanged taking a plain int (one of the
+// ClientState* constants), because gomobile cannot bind named types.
+type ConnectionListener interface {
+	OnStateChanged(state int)
+	OnConnected()
+	OnDisconnected()
+	OnConnecting()
+	OnDisconnecting()
+	OnAddressChanged(string, string)
+	OnPeersListChanged(int)
+}
+
+// connectionListenerAdapter adapts the gomobile-facing ConnectionListener to
+// peer.Listener, converting the typed state to the int the binding carries.
+type connectionListenerAdapter struct {
+	ConnectionListener
+}
+
+func (a connectionListenerAdapter) OnStateChanged(state peer.ClientState) {
+	a.ConnectionListener.OnStateChanged(int(state))
+}

client/android/login.go

@@ -4,8 +4,11 @@ import (
 	"context"
 	"fmt"
 
+	log "github.com/sirupsen/logrus"
+
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mobile"
 	"github.com/netbirdio/netbird/client/system"
 )
 
@@ -36,12 +39,20 @@ type Auth struct {
 }
 
 // NewAuth instantiate Auth struct and validate the management URL
+//
+// The configuration at cfgPath is reused when one is already there, and only created when it is
+// not. Building a fresh in-memory config unconditionally gives the client a new WireGuard key on
+// every call: the peer registers under that key, the key is written out, and any peer registered by
+// an earlier call is orphaned on the server. It also breaks a client that enrols and then runs from
+// the persisted config, because the identity it registered is not the one it runs with — the
+// management stream rejects it with "no peer auth method provided".
 func NewAuth(cfgPath string, mgmURL string) (*Auth, error) {
 	inputCfg := profilemanager.ConfigInput{
+		ConfigPath:    cfgPath,
 		ManagementURL: mgmURL,
 	}
 
-	cfg, err := profilemanager.CreateInMemoryConfig(inputCfg)
+	cfg, err := profilemanager.UpdateOrCreateConfig(inputCfg)
 	if err != nil {
 		return nil, err
 	}
@@ -53,11 +64,14 @@ func NewAuth(cfgPath string, mgmURL string) (*Auth, error) {
 	}, nil
 }
 
-// NewAuthWithConfig instantiate Auth based on existing config
-func NewAuthWithConfig(ctx context.Context, config *profilemanager.Config) *Auth {
+// NewAuthWithConfig instantiate Auth based on existing config. cfgPath is the
+// file the config was loaded from; it identifies the profile whose account email
+// backs the login_hint.
+func NewAuthWithConfig(ctx context.Context, config *profilemanager.Config, cfgPath string) *Auth {
 	return &Auth{
-		ctx:    ctx,
-		config: config,
+		ctx:     ctx,
+		config:  config,
+		cfgPath: cfgPath,
 	}
 }
 
@@ -150,12 +164,14 @@ func (a *Auth) login(urlOpener URLOpener, isAndroidTV bool) error {
 	}
 
 	jwtToken := ""
+	email := ""
 	if needsLogin {
 		tokenInfo, err := a.foregroundGetTokenInfo(authClient, urlOpener, isAndroidTV)
 		if err != nil {
 			return fmt.Errorf("interactive sso login failed: %v", err)
 		}
 		jwtToken = tokenInfo.GetTokenToUse()
+		email = tokenInfo.Email
 	}
 
 	err, _ = authClient.Login(a.ctx, "", jwtToken)
@@ -163,27 +179,62 @@ func (a *Auth) login(urlOpener URLOpener, isAndroidTV bool) error {
 		return fmt.Errorf("login failed: %v", err)
 	}
 
+	// Stored after Login, not before: a rejected token must not leave a hint
+	// pointing at an account that cannot be used.
+	if email != "" && a.cfgPath != "" {
+		if err := mobile.WriteProfileEmail(a.cfgPath, email); err != nil {
+			log.Warnf("failed to store profile account email: %v", err)
+		}
+	}
+
 	go urlOpener.OnLoginSuccess()
 
 	return nil
 }
 
 func (a *Auth) foregroundGetTokenInfo(authClient *auth.Auth, urlOpener URLOpener, isAndroidTV bool) (*auth.TokenInfo, error) {
-	oAuthFlow, err := authClient.GetOAuthFlow(a.ctx, isAndroidTV)
+	oAuthFlow, err := authClient.GetOAuthFlow(a.ctx, isAndroidTV, profileLoginHint(a.cfgPath))
 	if err != nil {
 		return nil, fmt.Errorf("failed to get OAuth flow: %v", err)
 	}
 
-	flowInfo, err := oAuthFlow.RequestAuthInfo(context.TODO())
+	return runOAuthFlow(a.ctx, oAuthFlow, urlOpener, nil)
+}
+
+// profileLoginHint returns the stored account email for the profile at cfgPath.
+// An empty hint is deliberate, not a fallback: a fresh profile leaves the
+// choice to the IdP. Switching accounts is done by switching or removing
+// profiles, not by logging out — logout keeps the email.
+func profileLoginHint(cfgPath string) string {
+	if cfgPath == "" {
+		return ""
+	}
+	return mobile.ReadProfileEmail(cfgPath)
+}
+
+// runOAuthFlow drives an already acquired OAuth flow to a token: requests the
+// flow info, presents the verification URL through the opener and waits for
+// the browser round-trip. Open is called synchronously — it is what marks the
+// surface as opened on the client side, and a fast token's OnLoginSuccess is
+// a no-op until it has, so the dismissal would be dropped rather than
+// delayed. Openers must therefore not block: they post their UI work and
+// return. onWaiting, when set, runs after the URL is shown, right before the
+// blocking wait.
+func runOAuthFlow(ctx context.Context, flow auth.OAuthFlow, urlOpener URLOpener, onWaiting func()) (*auth.TokenInfo, error) {
+	flowInfo, err := flow.RequestAuthInfo(ctx)
 	if err != nil {
-		return nil, fmt.Errorf("getting a request OAuth flow info failed: %v", err)
+		return nil, fmt.Errorf("request auth info: %w", err)
 	}
 
-	go urlOpener.Open(flowInfo.VerificationURIComplete, flowInfo.UserCode)
+	urlOpener.Open(flowInfo.VerificationURIComplete, flowInfo.UserCode)
 
-	tokenInfo, err := oAuthFlow.WaitToken(a.ctx, flowInfo)
+	if onWaiting != nil {
+		onWaiting()
+	}
+
+	tokenInfo, err := flow.WaitToken(ctx, flowInfo)
 	if err != nil {
-		return nil, fmt.Errorf("waiting for browser login failed: %v", err)
+		return nil, fmt.Errorf("wait for token: %w", err)
 	}
 
 	return &tokenInfo, nil

client/android/login_test.go

@@ -0,0 +1,51 @@
+package android
+
+import (
+	"path/filepath"
+	"testing"
+)
+
+// NewAuth must reuse the configuration already at cfgPath rather than building a fresh one.
+//
+// Creating a new in-memory config on every call gives the client a new WireGuard private key each
+// time. The peer registers under that key and the key is written out, so a peer registered by an
+// earlier call is orphaned on the server — a client that enrols twice leaves two entries and owns
+// neither. It also breaks enrol-then-run: RunWithoutLogin reloads the configuration from disk, so
+// the identity that registered is not the identity that runs, and the management stream rejects it
+// with "no peer auth method provided, please use a setup key or interactive SSO login".
+func TestNewAuth_ReusesPersistedIdentity(t *testing.T) {
+	cfgPath := filepath.Join(t.TempDir(), "config.json")
+
+	first, err := NewAuth(cfgPath, "https://api.example.com:443")
+	if err != nil {
+		t.Fatalf("first NewAuth: %v", err)
+	}
+	if first.config.PrivateKey == "" {
+		t.Fatal("first NewAuth produced no private key")
+	}
+
+	second, err := NewAuth(cfgPath, "https://api.example.com:443")
+	if err != nil {
+		t.Fatalf("second NewAuth: %v", err)
+	}
+
+	if second.config.PrivateKey != first.config.PrivateKey {
+		t.Errorf("private key changed between calls: a second enrolment would orphan the peer registered by the first")
+	}
+}
+
+// A missing configuration is still created, so a first enrolment works unchanged.
+func TestNewAuth_CreatesConfigWhenAbsent(t *testing.T) {
+	cfgPath := filepath.Join(t.TempDir(), "config.json")
+
+	auth, err := NewAuth(cfgPath, "https://api.example.com:443")
+	if err != nil {
+		t.Fatalf("NewAuth: %v", err)
+	}
+	if auth.config == nil || auth.config.PrivateKey == "" {
+		t.Fatal("NewAuth did not create a usable configuration")
+	}
+	if auth.cfgPath != cfgPath {
+		t.Errorf("cfgPath = %q, want %q", auth.cfgPath, cfgPath)
+	}
+}

client/android/profile_manager.go

@@ -3,42 +3,37 @@
 package android
 
 import (
-	"fmt"
-	"os"
-	"path/filepath"
-
-	log "github.com/sirupsen/logrus"
-
-	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/mobile"
 )
 
 const (
-	// Android-specific config filename (different from desktop default.json)
-	defaultConfigFilename = "netbird.cfg"
-	// Subdirectory for non-default profiles (must match Java Preferences.java)
-	profilesSubdir = "profiles"
-	// Android uses a single user context per app (non-empty username required by ServiceManager)
+	// Android uses a single user context per app.
 	androidUsername = "android"
 )
 
-// Profile represents a profile for gomobile
+// Profile represents a profile for gomobile.
 type Profile struct {
-	ID       string
-	Name     string
+	ID   string
+	Name string
+	// Email is the account this profile last logged in with, "" if it never
+	// completed an SSO login. Kept across logouts; cleared when the profile is
+	// removed. See client/mobile/profile_state.go.
+	Email    string
 	IsActive bool
 }
 
-// ProfileArray wraps profiles for gomobile compatibility
+// ProfileArray wraps profiles for gomobile compatibility (gomobile cannot
+// bind Go slices directly).
 type ProfileArray struct {
 	items []*Profile
 }
 
-// Length returns the number of profiles
+// Length returns the number of profiles.
 func (p *ProfileArray) Length() int {
 	return len(p.items)
 }
 
-// Get returns the profile at index i
+// Get returns the profile at index i, or nil if out of range.
 func (p *ProfileArray) Get(i int) *Profile {
 	if i < 0 || i >= len(p.items) {
 		return nil
@@ -46,227 +41,98 @@ func (p *ProfileArray) Get(i int) *Profile {
 	return p.items[i]
 }
 
-/*
-
-/data/data/io.netbird.client/files/           ← configDir parameter
-├── netbird.cfg                                 ← Default profile config
-├── state.json                                  ← Default profile state
-├── active_profile.json                         ← Active profile tracker (JSON with Name + Username)
-└── profiles/                                   ← Subdirectory for non-default profiles
-    ├── work.json                              			← Legacy work profile config
-    ├── work.state.json                        			← Legacy work profile state
-    ├── 4c5f5c8198c3989cffb5b5394f5a7ae0.json  			← ID profile config
-    ├── 4c5f5c8198c3989cffb5b5394f5a7ae0.state.json ← ID profile state
-*/
-
-// ProfileManager manages profiles for Android
-// It wraps the internal profilemanager to provide Android-specific behavior
+// ProfileManager adapts the shared mobile profile manager (client/mobile) to
+// gomobile-friendly types. See that package for the on-disk layout and
+// semantics.
 type ProfileManager struct {
-	configDir  string
-	serviceMgr *profilemanager.ServiceManager
+	impl *mobile.ProfileManager
 }
 
-// NewProfileManager creates a new profile manager for Android
+// NewProfileManager creates a new profile manager for Android. configDir is
+// the app's files directory.
 func NewProfileManager(configDir string) *ProfileManager {
-	// Set the default config path for Android (stored in root configDir, not profiles/)
-	defaultConfigPath := filepath.Join(configDir, defaultConfigFilename)
-
-	// Set global paths for Android
-	profilemanager.DefaultConfigPathDir = configDir
-	profilemanager.DefaultConfigPath = defaultConfigPath
-	profilemanager.ActiveProfileStatePath = filepath.Join(configDir, "active_profile.json")
-
-	// Create ServiceManager with profiles/ subdirectory
-	// This avoids modifying the global ConfigDirOverride for profile listing
-	profilesDir := filepath.Join(configDir, profilesSubdir)
-	serviceMgr := profilemanager.NewServiceManagerWithProfilesDir(defaultConfigPath, profilesDir)
-
-	return &ProfileManager{
-		configDir:  configDir,
-		serviceMgr: serviceMgr,
-	}
+	return &ProfileManager{impl: mobile.NewProfileManager(configDir, androidUsername)}
 }
 
-// ListProfiles returns all available profiles
+// ListProfiles returns all available profiles, including the default profile,
+// with their active status set.
 func (pm *ProfileManager) ListProfiles() (*ProfileArray, error) {
-	// Use ServiceManager (looks in profiles/ directory, checks active_profile.json for IsActive)
-	internalProfiles, err := pm.serviceMgr.ListProfiles(androidUsername)
+	profiles, err := pm.impl.ListProfiles()
 	if err != nil {
-		return nil, fmt.Errorf("failed to list profiles: %w", err)
+		return nil, err
 	}
 
-	// Convert internal profiles to Android Profile type
-	var profiles []*Profile
-	for _, p := range internalProfiles {
-		profiles = append(profiles, &Profile{
-			ID:       p.ID.String(),
-			Name:     p.Name,
-			IsActive: p.IsActive,
-		})
+	items := make([]*Profile, 0, len(profiles))
+	for i := range profiles {
+		items = append(items, fromMobileProfile(&profiles[i]))
 	}
-
-	return &ProfileArray{items: profiles}, nil
+	return &ProfileArray{items: items}, nil
 }
 
-// GetActiveProfile returns the currently active profile name
+// GetActiveProfile returns the currently active profile.
 func (pm *ProfileManager) GetActiveProfile() (*Profile, error) {
-	// Use ServiceManager to stay consistent with ListProfiles
-	// ServiceManager uses active_profile.json
-	activeState, err := pm.serviceMgr.GetActiveProfileState()
+	p, err := pm.impl.GetActiveProfile()
 	if err != nil {
-		return nil, fmt.Errorf("failed to get active profile: %w", err)
+		return nil, err
 	}
-
-	// ActiveProfileState only stores the ID (and username), not the display
-	// name. Resolve the ID to the full profile so callers get the real Name.
-	prof, err := pm.serviceMgr.ResolveProfile(activeState.ID.String(), androidUsername)
-	if err != nil {
-		return nil, fmt.Errorf("failed to resolve active profile %q: %w", activeState.ID, err)
-	}
-	return &Profile{ID: prof.ID.String(), Name: prof.Name, IsActive: true}, nil
+	return fromMobileProfile(p), nil
 }
 
-// SwitchProfile switches to a different profile
+// SwitchProfile records the given profile ID as the active profile. The caller
+// must stop the VPN tunnel before switching.
 func (pm *ProfileManager) SwitchProfile(id string) error {
-	// Use ServiceManager to stay consistent with ListProfiles
-	// ServiceManager uses active_profile.json
-	err := pm.serviceMgr.SetActiveProfileState(&profilemanager.ActiveProfileState{
-		ID:       profilemanager.ID(id),
-		Username: androidUsername,
-	})
-	if err != nil {
-		return fmt.Errorf("failed to switch profile: %w", err)
-	}
-
-	log.Infof("switched to profile: %s", id)
-	return nil
+	return pm.impl.SwitchProfile(id)
 }
 
-// AddProfile creates a new profile
+// AddProfile creates a new profile with the given display name and a
+// generated ID.
 func (pm *ProfileManager) AddProfile(profileName string) error {
-	// Use ServiceManager (creates profile in profiles/ directory)
-	profile, err := pm.serviceMgr.AddProfile(profileName, androidUsername)
-	if err != nil {
-		return fmt.Errorf("failed to add profile: %w", err)
-	}
-
-	log.Infof("created new profile: %s", profile.ID)
-	return nil
+	_, err := pm.impl.AddProfile(profileName)
+	return err
 }
 
-// LogoutProfile logs out from a profile (clears authentication)
-func (pm *ProfileManager) LogoutProfile(id string) error {
-	configPath, err := pm.getProfileConfigPath(id)
-	if err != nil {
-		return err
-	}
-
-	if !profilemanager.IsValidProfileFilenameStem(profilemanager.ID(id)) {
-		return fmt.Errorf("id '%s' is not valid", id)
-	}
-
-	// Check if profile exists
-	if _, err := os.Stat(configPath); os.IsNotExist(err) {
-		return fmt.Errorf("profile '%s' does not exist", id)
-	}
-
-	// Read current config using internal profilemanager
-	config, err := profilemanager.ReadConfig(configPath)
-	if err != nil {
-		return fmt.Errorf("failed to read profile config: %w", err)
-	}
-
-	// Clear authentication by removing private key and SSH key
-	config.PrivateKey = ""
-	config.SSHKey = ""
-
-	// Save config using internal profilemanager
-	if err := profilemanager.WriteOutConfig(configPath, config); err != nil {
-		return fmt.Errorf("failed to save config: %w", err)
-	}
-
-	log.Infof("logged out from profile: %s", id)
-	return nil
-}
-
-// RenameProfile changes a profile's display name. The profile ID, and therefore
-// its on-disk filename, is left untouched: only the "name" field of the config
-// is rewritten. This works for the default profile too, whose config lives in
-// netbird.cfg rather than under profiles/.
+// RenameProfile changes the display name of the profile identified by id. The
+// on-disk filename (the ID) is left unchanged.
 func (pm *ProfileManager) RenameProfile(id string, newName string) error {
-	if err := pm.serviceMgr.RenameProfile(profilemanager.ID(id), androidUsername, newName); err != nil {
-		return fmt.Errorf("failed to rename profile: %w", err)
-	}
-
-	log.Infof("renamed profile %s to: %s", id, newName)
-	return nil
+	return pm.impl.RenameProfile(id, newName)
 }
 
-// RemoveProfile deletes a profile
+// LogoutProfile clears authentication data for a profile, forcing a re-login.
+// The management URL and other settings are preserved.
+func (pm *ProfileManager) LogoutProfile(id string) error {
+	return pm.impl.LogoutProfile(id)
+}
+
+// RemoveProfile deletes a profile. The default profile and the active profile
+// cannot be removed.
 func (pm *ProfileManager) RemoveProfile(id string) error {
-	// Use ServiceManager (removes profile from profiles/ directory)
-	if err := pm.serviceMgr.RemoveProfile(profilemanager.ID(id), androidUsername); err != nil {
-		return fmt.Errorf("failed to remove profile: %w", err)
-	}
-
-	log.Infof("removed profile: %s", id)
-	return nil
+	return pm.impl.RemoveProfile(id)
 }
 
-// getProfileConfigPath returns the config file path for a profile
-// This is needed for Android-specific path handling (netbird.cfg for default profile)
-func (pm *ProfileManager) getProfileConfigPath(id string) (string, error) {
-	if !profilemanager.IsValidProfileFilenameStem(profilemanager.ID(id)) {
-		return "", fmt.Errorf("id %q is not valid", id)
-	}
-
-	if id == profilemanager.DefaultProfileName {
-		// Android uses netbird.cfg for default profile instead of default.json
-		// Default profile is stored in root configDir, not in profiles/
-		return filepath.Join(pm.configDir, defaultConfigFilename), nil
-	}
-
-	profilesDir := filepath.Join(pm.configDir, profilesSubdir)
-	return filepath.Join(profilesDir, id+".json"), nil
-}
-
-// GetConfigPath returns the config file path for a given profile id
-// Java should call this instead of constructing paths with Preferences.configFile()
+// GetConfigPath returns the config file path for the given profile ID. Java
+// should call this instead of constructing paths with Preferences.configFile().
 func (pm *ProfileManager) GetConfigPath(id string) (string, error) {
-	return pm.getProfileConfigPath(id)
+	return pm.impl.GetConfigPath(id)
 }
 
-// GetStateFilePath returns the state file path for a given profile
-// Java should call this instead of constructing paths with Preferences.stateFile()
+// GetStateFilePath returns the state file path for the given profile ID. Java
+// should call this instead of constructing paths with Preferences.stateFile().
 func (pm *ProfileManager) GetStateFilePath(id string) (string, error) {
-	if id == "" || id == profilemanager.DefaultProfileName {
-		return filepath.Join(pm.configDir, "state.json"), nil
-	}
-
-	if !profilemanager.IsValidProfileFilenameStem(profilemanager.ID(id)) {
-		return "", fmt.Errorf("id %q is not valid", id)
-	}
-
-	profilesDir := filepath.Join(pm.configDir, profilesSubdir)
-	return filepath.Join(profilesDir, id+".state.json"), nil
+	return pm.impl.GetStateFilePath(id)
 }
 
-// GetActiveConfigPath returns the config file path for the currently active profile
-// Java should call this instead of Preferences.getActiveProfileName() + Preferences.configFile()
+// GetActiveConfigPath returns the config file path for the currently active
+// profile.
 func (pm *ProfileManager) GetActiveConfigPath() (string, error) {
-	activeProfile, err := pm.GetActiveProfile()
-	if err != nil {
-		return "", fmt.Errorf("failed to get active profile: %w", err)
-	}
-	return pm.GetConfigPath(activeProfile.ID)
+	return pm.impl.GetActiveConfigPath()
 }
 
-// GetActiveStateFilePath returns the state file path for the currently active profile
-// Java should call this instead of Preferences.getActiveProfileName() + Preferences.stateFile()
+// GetActiveStateFilePath returns the state file path for the currently active
+// profile.
 func (pm *ProfileManager) GetActiveStateFilePath() (string, error) {
-	activeProfile, err := pm.GetActiveProfile()
-	if err != nil {
-		return "", fmt.Errorf("failed to get active profile: %w", err)
-	}
-	return pm.GetStateFilePath(activeProfile.ID)
+	return pm.impl.GetActiveStateFilePath()
+}
+
+func fromMobileProfile(p *mobile.Profile) *Profile {
+	return &Profile{ID: p.ID, Name: p.Name, Email: p.Email, IsActive: p.IsActive}
 }

client/android/profile_prefs.go

@@ -0,0 +1,37 @@
+//go:build android
+
+package android
+
+import (
+	"fmt"
+
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+)
+
+type prefsStore interface {
+	Get(namespace string, v any) (bool, error)
+	Put(namespace string, v any) error
+}
+
+type profilePrefs struct {
+	prefs *profilemanager.Prefs
+}
+
+func newProfilePrefs(configDir, profileID string) (*profilePrefs, error) {
+	if configDir == "" || profileID == "" {
+		return nil, fmt.Errorf("profile prefs require a config dir and profile ID")
+	}
+	prefs, err := NewProfileManager(configDir).impl.ProfilePrefs(profileID)
+	if err != nil {
+		return nil, err
+	}
+	return &profilePrefs{prefs: prefs}, nil
+}
+
+func (p *profilePrefs) Get(namespace string, v any) (bool, error) {
+	return p.prefs.Get(namespace, v)
+}
+
+func (p *profilePrefs) Put(namespace string, v any) error {
+	return p.prefs.Put(namespace, v)
+}

client/android/session.go

@@ -278,7 +278,7 @@ func (c *Client) endExtend() {
 }
 
 func (c *Client) extendAuthSession(ctx context.Context, urlOpener URLOpener, isAndroidTV bool) error {
-	cfg, _, cc := c.stateSnapshot()
+	cfg, cfgPath, cc := c.authSnapshot()
 	if cfg == nil || cc == nil {
 		return fmt.Errorf("engine is not running")
 	}
@@ -293,7 +293,10 @@ func (c *Client) extendAuthSession(ctx context.Context, urlOpener URLOpener, isA
 	}
 	defer authClient.Close()
 
-	a := &Auth{ctx: ctx, config: cfg}
+	// Passing the config path makes the flow pick up the login_hint: an extend
+	// renews the session of the account already signed in, so it must not stop to
+	// offer a choice.
+	a := NewAuthWithConfig(ctx, cfg, cfgPath)
 	tokenInfo, err := a.foregroundGetTokenInfo(authClient, urlOpener, isAndroidTV)
 	if err != nil {
 		return fmt.Errorf("interactive sso login failed: %v", err)

client/android/split_tunnel.go

@@ -0,0 +1,125 @@
+package android
+
+// SplitTunnelMode is which of the two selections, if either, the tunnel applies.
+// Its values land in the profile's stored preferences, so the constants below
+// are append-only and must never be reordered.
+type SplitTunnelMode int
+
+const (
+	modeOff SplitTunnelMode = iota
+	modeExclude
+	modeInclude
+)
+
+// The same modes as basic ints. gomobile drops a constant whose type is not a
+// basic one, so these are what reaches the generated Java bindings, and they
+// keep the Android side tied to the values above instead of repeating 0, 1, 2.
+const (
+	SplitTunnelModeOff     = int(modeOff)
+	SplitTunnelModeExclude = int(modeExclude)
+	SplitTunnelModeInclude = int(modeInclude)
+)
+
+type splitTunnelSection struct {
+	Mode     SplitTunnelMode `json:"mode"`
+	Excluded []string        `json:"excluded"`
+	Included []string        `json:"included"`
+}
+
+// PackageList wraps []string for gomobile compatibility.
+type PackageList struct {
+	items []string
+}
+
+// NewPackageList creates an empty list to fill via Add.
+func NewPackageList() *PackageList {
+	return &PackageList{}
+}
+
+// Add appends a package name, ignoring empty ones.
+func (l *PackageList) Add(s string) {
+	if s == "" {
+		return
+	}
+	l.items = append(l.items, s)
+}
+
+// Size returns the number of entries.
+func (l *PackageList) Size() int {
+	return len(l.items)
+}
+
+// Get returns the entry at index i, or an empty string when out of range.
+func (l *PackageList) Get(i int) string {
+	if i < 0 || i >= len(l.items) {
+		return ""
+	}
+	return l.items[i]
+}
+
+// SplitTunnelSettings is one profile's choice of which applications the tunnel
+// carries. The two selections are kept apart because the platform applies one
+// or the other and never both, and so that switching mode does not throw away
+// the picks made in the other one.
+//
+// Mode is an int rather than a SplitTunnelMode because gomobile carries only
+// basic types across the binding. It holds one of the SplitTunnelMode*
+// constants.
+type SplitTunnelSettings struct {
+	Mode     int
+	Excluded *PackageList
+	Included *PackageList
+}
+
+// NewSplitTunnelSettings creates settings that carry every application.
+func NewSplitTunnelSettings() *SplitTunnelSettings {
+	return &SplitTunnelSettings{
+		Mode:     SplitTunnelModeOff,
+		Excluded: NewPackageList(),
+		Included: NewPackageList(),
+	}
+}
+
+func packagesOf(list *PackageList) []string {
+	if list == nil {
+		return nil
+	}
+	out := make([]string, 0, len(list.items))
+	out = append(out, list.items...)
+	return out
+}
+
+// normalizeSplitTunnelMode maps anything outside the known set to off, so a mode
+// written by a newer build degrades to carrying every application rather than to
+// some other mode's behaviour.
+func normalizeSplitTunnelMode(mode SplitTunnelMode) SplitTunnelMode {
+	switch mode {
+	case modeExclude, modeInclude:
+		return mode
+	default:
+		return modeOff
+	}
+}
+
+func settingsFromSection(section splitTunnelSection) *SplitTunnelSettings {
+	out := NewSplitTunnelSettings()
+	out.Mode = int(normalizeSplitTunnelMode(section.Mode))
+	for _, pkg := range section.Excluded {
+		out.Excluded.Add(pkg)
+	}
+	for _, pkg := range section.Included {
+		out.Included.Add(pkg)
+	}
+	return out
+}
+
+func sectionFromSettings(settings *SplitTunnelSettings) splitTunnelSection {
+	if settings == nil {
+		settings = NewSplitTunnelSettings()
+	}
+	return splitTunnelSection{
+		Mode:     normalizeSplitTunnelMode(SplitTunnelMode(settings.Mode)),
+		Excluded: packagesOf(settings.Excluded),
+		Included: packagesOf(settings.Included),
+	}
+}

client/android/split_tunnel_store.go

@@ -0,0 +1,34 @@
+//go:build android
+
+package android
+
+const splitTunnelNamespace = "split-tunnel"
+
+// SplitTunnelStore reads and writes a profile's split tunnelling settings.
+type SplitTunnelStore struct {
+	prefs prefsStore
+}
+
+// NewSplitTunnelStore opens the split tunnelling store of the given profile.
+func NewSplitTunnelStore(configDir, profileID string) (*SplitTunnelStore, error) {
+	prefs, err := newProfilePrefs(configDir, profileID)
+	if err != nil {
+		return nil, err
+	}
+	return &SplitTunnelStore{prefs: prefs}, nil
+}
+
+// Load returns the stored settings, or settings that carry every application
+// when the profile has none saved.
+func (s *SplitTunnelStore) Load() (*SplitTunnelSettings, error) {
+	var section splitTunnelSection
+	if _, err := s.prefs.Get(splitTunnelNamespace, &section); err != nil {
+		return nil, err
+	}
+	return settingsFromSection(section), nil
+}
+
+// Save replaces the stored settings.
+func (s *SplitTunnelStore) Save(settings *SplitTunnelSettings) error {
+	return s.prefs.Put(splitTunnelNamespace, sectionFromSettings(settings))
+}

client/android/split_tunnel_test.go

@@ -0,0 +1,151 @@
+package android
+
+import (
+	"encoding/json"
+	"reflect"
+	"testing"
+)
+
+func TestNormalizeSplitTunnelMode(t *testing.T) {
+	tests := []struct {
+		name string
+		mode SplitTunnelMode
+		want SplitTunnelMode
+	}{
+		{name: "exclude is kept", mode: modeExclude, want: modeExclude},
+		{name: "include is kept", mode: modeInclude, want: modeInclude},
+		{name: "off is kept", mode: modeOff, want: modeOff},
+		{name: "a mode from a newer build falls back to off", mode: SplitTunnelMode(7), want: modeOff},
+		{name: "a negative mode falls back to off", mode: SplitTunnelMode(-1), want: modeOff},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := normalizeSplitTunnelMode(tt.mode); got != tt.want {
+				t.Errorf("normalizeSplitTunnelMode(%d) = %d, want %d", tt.mode, got, tt.want)
+			}
+		})
+	}
+}
+
+// The constants the Android side reads must stay the values the store writes:
+// gomobile carries the ints below, not the typed constants they mirror.
+func TestSplitTunnelModeConstantsMirrorTheTypedOnes(t *testing.T) {
+	if SplitTunnelModeOff != int(modeOff) {
+		t.Errorf("off = %d, want %d", SplitTunnelModeOff, modeOff)
+	}
+	if SplitTunnelModeExclude != int(modeExclude) {
+		t.Errorf("exclude = %d, want %d", SplitTunnelModeExclude, modeExclude)
+	}
+	if SplitTunnelModeInclude != int(modeInclude) {
+		t.Errorf("include = %d, want %d", SplitTunnelModeInclude, modeInclude)
+	}
+}
+
+func TestSettingsFromSection(t *testing.T) {
+	got := settingsFromSection(splitTunnelSection{
+		Mode:     modeExclude,
+		Excluded: []string{"com.example.a", "com.example.b"},
+		Included: []string{"com.example.c"},
+	})
+
+	if got.Mode != SplitTunnelModeExclude {
+		t.Errorf("mode = %d, want %d", got.Mode, SplitTunnelModeExclude)
+	}
+	if got.Excluded.Size() != 2 || got.Excluded.Get(0) != "com.example.a" {
+		t.Errorf("excluded = %v, want the two stored packages", packagesOf(got.Excluded))
+	}
+	if got.Included.Size() != 1 || got.Included.Get(0) != "com.example.c" {
+		t.Errorf("included = %v, want the stored package", packagesOf(got.Included))
+	}
+}
+
+// A profile that has never stored anything decodes into an empty section, and
+// must come back as settings that carry every application rather than as nil
+// lists the caller would have to guard against.
+func TestSettingsFromEmptySectionCarriesEverything(t *testing.T) {
+	got := settingsFromSection(splitTunnelSection{})
+
+	if got.Mode != SplitTunnelModeOff {
+		t.Errorf("mode = %d, want %d", got.Mode, SplitTunnelModeOff)
+	}
+	if got.Excluded == nil || got.Included == nil {
+		t.Fatal("both selections must be usable lists, not nil")
+	}
+	if got.Excluded.Size() != 0 || got.Included.Size() != 0 {
+		t.Errorf("selections = %v/%v, want both empty", packagesOf(got.Excluded), packagesOf(got.Included))
+	}
+}
+
+// The section is what the profile's preference file holds, so the mode has to
+// survive a JSON round trip as the number the constants name.
+func TestSectionEncodesTheModeAsItsNumber(t *testing.T) {
+	raw, err := json.Marshal(sectionFromSettings(&SplitTunnelSettings{Mode: SplitTunnelModeInclude}))
+	if err != nil {
+		t.Fatalf("marshal section: %v", err)
+	}
+
+	var back splitTunnelSection
+	if err := json.Unmarshal(raw, &back); err != nil {
+		t.Fatalf("unmarshal section: %v", err)
+	}
+	if back.Mode != modeInclude {
+		t.Errorf("mode = %d, want %d, from %s", back.Mode, modeInclude, raw)
+	}
+}
+
+func TestSectionFromSettingsRoundTrip(t *testing.T) {
+	settings := NewSplitTunnelSettings()
+	settings.Mode = SplitTunnelModeInclude
+	settings.Included.Add("com.example.a")
+	settings.Excluded.Add("com.example.b")
+
+	section := sectionFromSettings(settings)
+	back := settingsFromSection(section)
+
+	if back.Mode != SplitTunnelModeInclude {
+		t.Errorf("mode = %d, want %d", back.Mode, SplitTunnelModeInclude)
+	}
+	if !reflect.DeepEqual(packagesOf(back.Included), []string{"com.example.a"}) {
+		t.Errorf("included = %v, want [com.example.a]", packagesOf(back.Included))
+	}
+	// The inactive selection survives, so switching mode back does not make the
+	// user pick their applications again.
+	if !reflect.DeepEqual(packagesOf(back.Excluded), []string{"com.example.b"}) {
+		t.Errorf("excluded = %v, want [com.example.b]", packagesOf(back.Excluded))
+	}
+}
+
+// A mode the Java side never sets, such as one left by a newer build, must not
+// reach the stored section either.
+func TestSectionFromSettingsNormalizesAnUnknownMode(t *testing.T) {
+	section := sectionFromSettings(&SplitTunnelSettings{Mode: 7})
+
+	if section.Mode != modeOff {
+		t.Errorf("mode = %d, want %d", section.Mode, modeOff)
+	}
+}
+
+func TestSectionFromNilSettings(t *testing.T) {
+	section := sectionFromSettings(nil)
+
+	if section.Mode != modeOff {
+		t.Errorf("mode = %d, want %d", section.Mode, modeOff)
+	}
+	if len(section.Excluded) != 0 || len(section.Included) != 0 {
+		t.Errorf("selections = %v/%v, want both empty", section.Excluded, section.Included)
+	}
+}
+
+func TestPackageListIgnoresEmptyAndBounds(t *testing.T) {
+	list := NewPackageList()
+	list.Add("com.example.a")
+	list.Add("")
+
+	if list.Size() != 1 {
+		t.Errorf("size = %d, want 1", list.Size())
+	}
+	if list.Get(-1) != "" || list.Get(5) != "" {
+		t.Error("out of range access must return an empty string")
+	}
+}

client/android/ssh_client.go

@@ -0,0 +1,649 @@
+//go:build android
+
+package android
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"strconv"
+	"strings"
+	"sync"
+	"time"
+
+	log "github.com/sirupsen/logrus"
+	gossh "golang.org/x/crypto/ssh"
+
+	"github.com/netbirdio/netbird/client/internal"
+	"github.com/netbirdio/netbird/client/internal/auth"
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	nbssh "github.com/netbirdio/netbird/client/ssh"
+	"github.com/netbirdio/netbird/client/ssh/detection"
+)
+
+const (
+	sshDialTimeout      = 30 * time.Second
+	sshDetectionTimeout = 5 * time.Second
+)
+
+// PasswordRequiredMarker tells Java to prompt for a password and retry. It is
+// a string because gomobile flattens errors to their message, so a sentinel
+// value would not survive the binding.
+const PasswordRequiredMarker = "netbird-ssh-password-required"
+
+// HostKeyUnknownMarker tells Java to show the fingerprint and, on confirmation,
+// retry with TrustHostKey set. The presented fingerprint is appended after the
+// marker so the prompt can display it and the retry can guard against a key
+// that changed between the two connects. Only regular (non-NetBird) servers
+// reach this: NetBird peers verify against the registry.
+const HostKeyUnknownMarker = "netbird-ssh-hostkey-unknown"
+
+var (
+	errPasswordRequired = errors.New(PasswordRequiredMarker)
+	errClientClosed     = errors.New("ssh client closed")
+)
+
+// errHostKeyUnknown carries the presented fingerprint so Connect can build the
+// marker message the Java side parses.
+type errHostKeyUnknown struct {
+	fingerprint string
+}
+
+func (e *errHostKeyUnknown) Error() string {
+	return HostKeyUnknownMarker + ":" + e.fingerprint
+}
+
+// SSHTerminalListener receives SSH session events. It is implemented in Java.
+//
+// All callbacks are invoked from goroutines and may run concurrently with each
+// other; the implementation must be safe to call from any thread.
+type SSHTerminalListener interface {
+	OnConnected()
+	OnData(data []byte)
+	OnClose(reason string)
+	OnError(message string)
+}
+
+// SSHClient is a NetBird-aware SSH client exposed to Java via gomobile.
+//
+// It dials through the running NetBird tunnel and runs a standard SSH session
+// on top with PTY enabled. Host-key verification uses the NetBird-provided
+// peer SSH host keys, identical to the desktop client.
+type SSHClient struct {
+	nb        *Client
+	mu        sync.Mutex
+	listener  SSHTerminalListener
+	urlOpener URLOpener
+
+	sshClient *gossh.Client
+	session   *gossh.Session
+	stdin     io.WriteCloser
+	closed    bool
+
+	// gen identifies the current connection attempt. Connect and Close bump it,
+	// so an in-flight dial or a reader left over from a previous connection
+	// finds itself stale and stays silent instead of publishing OnConnected or
+	// OnClose for a connection the caller already abandoned.
+	gen        uint64
+	dialCancel context.CancelFunc
+
+	// knownHostsConfigDir and knownHostsProfile locate the TOFU store for
+	// regular SSH servers in the profile's preferences. Java supplies them,
+	// since an overlay IP is a different host under a different profile. Empty
+	// until set: without them a regular server cannot be verified and Connect
+	// refuses one.
+	knownHostsConfigDir string
+	knownHostsProfile   string
+	// trustHostKey carries the fingerprint the user confirmed on a previous
+	// attempt, so the retry accepts exactly that key and persists it.
+	trustHostKey string
+}
+
+// NewSSHClient creates a new SSH client bound to the running NetBird Client.
+func NewSSHClient(c *Client) *SSHClient {
+	return &SSHClient{nb: c}
+}
+
+// SetListener registers the Java listener. Must be called before Connect to
+// receive any events.
+func (s *SSHClient) SetListener(l SSHTerminalListener) {
+	s.mu.Lock()
+	s.listener = l
+	s.mu.Unlock()
+}
+
+// SetURLOpener registers the Java URL opener used to display the device-code
+// authorization page in a Custom Tabs window when the target peer requires
+// JWT authentication. Must be set before Connect to be effective.
+func (s *SSHClient) SetURLOpener(opener URLOpener) {
+	s.mu.Lock()
+	s.urlOpener = opener
+	s.mu.Unlock()
+}
+
+// SetKnownHostsStore points the TOFU host-key store at a profile's preferences.
+// Must be set before connecting to a regular SSH server; without it such a
+// server cannot be verified and Connect refuses one.
+func (s *SSHClient) SetKnownHostsStore(configDir, profileID string) {
+	s.mu.Lock()
+	s.knownHostsConfigDir = configDir
+	s.knownHostsProfile = profileID
+	s.mu.Unlock()
+}
+
+// TrustHostKey records the fingerprint the user confirmed for a regular server,
+// so the next Connect accepts that exact key and adds it to the known-hosts
+// store. Passing a fingerprint that no longer matches makes the connect fail
+// rather than trust a key that changed since the prompt.
+func (s *SSHClient) TrustHostKey(fingerprint string) {
+	s.mu.Lock()
+	s.trustHostKey = fingerprint
+	s.mu.Unlock()
+}
+
+// Connect dials the SSH server through the NetBird tunnel and performs the
+// SSH handshake. It auto-detects the server type via SSH banner inspection
+// and selects the appropriate authentication path:
+//
+//   - NetBird-SSH server requiring JWT: launches the OAuth 2.0 device-code
+//     flow, opens the verification URL through the registered URLOpener, and
+//     uses the resulting token as the SSH password. Host-key verification
+//     uses the NetBird peer registry.
+//   - NetBird-SSH server without JWT: authenticates with the NetBird SSH
+//     private key. Host-key verification uses the NetBird peer registry.
+//   - Regular SSH server (e.g. OpenSSH): authenticates with the NetBird key
+//     first (so a user-installed NetBird public key works), then falls back
+//     to the supplied password if non-empty. Host-key verification is
+//     trust-on-first-use against the per-profile known-hosts store.
+//
+// The password parameter is only consulted for regular SSH servers.
+func (s *SSHClient) Connect(host string, port int, user, password string) error {
+	if port < 1 || port > 65535 {
+		return fmt.Errorf("invalid port: %d", port)
+	}
+
+	cfg, cfgPath, cc := s.nb.authSnapshot()
+	if cc == nil {
+		return errors.New("netbird client not running")
+	}
+	if cfg == nil {
+		return errors.New("netbird config not loaded")
+	}
+	engine := cc.Engine()
+	if engine == nil {
+		return errors.New("netbird engine not available")
+	}
+
+	s.mu.Lock()
+	s.gen++
+	gen := s.gen
+	s.mu.Unlock()
+
+	serverType := detectServerType(host, port)
+	log.Debugf("SSH server type: %s", serverType)
+
+	authMethods, hostKeyCallback, err := s.buildAuth(cfg, cfgPath, engine, serverType, password)
+	if err != nil {
+		return err
+	}
+
+	clientConfig := &gossh.ClientConfig{
+		User:            user,
+		Auth:            authMethods,
+		HostKeyCallback: hostKeyCallback,
+		Timeout:         sshDialTimeout,
+	}
+	err = s.dialAndHandshake(gen, host, port, clientConfig)
+
+	// An unknown host key is a prompt, not a failure: return the marker intact
+	// (rootCause would unwrap it) so Java can show the fingerprint and retry.
+	var unknownHost *errHostKeyUnknown
+	if errors.As(err, &unknownHost) {
+		return errors.New(unknownHost.Error())
+	}
+
+	// A regular server may still accept a password, so let the caller ask for
+	// one instead of failing. NetBird servers never use a password, so a
+	// failure there is genuine.
+	if err != nil && serverType != detection.ServerTypeNetBirdJWT &&
+		serverType != detection.ServerTypeNetBirdNoJWT && isAuthFailure(err) &&
+		passwordCouldHelp(err, password != "") {
+		return errPasswordRequired
+	}
+	if err != nil {
+		return rootCause(err)
+	}
+	return nil
+}
+
+// StartSession requests a PTY and starts an interactive shell. Output from
+// the session is forwarded to the listener via OnData.
+func (s *SSHClient) StartSession(cols, rows int) error {
+	err := s.startSession(cols, rows)
+	if err != nil {
+		log.Infof("SSH: start session failed: %v", err)
+		return rootCause(err)
+	}
+	return nil
+}
+
+// Write sends data to the SSH session stdin.
+func (s *SSHClient) Write(data []byte) error {
+	s.mu.Lock()
+	stdin := s.stdin
+	s.mu.Unlock()
+	if stdin == nil {
+		return errors.New("ssh session not started")
+	}
+	if _, err := stdin.Write(data); err != nil {
+		return fmt.Errorf("write stdin: %w", err)
+	}
+	return nil
+}
+
+// Resize updates the PTY window size.
+func (s *SSHClient) Resize(cols, rows int) error {
+	s.mu.Lock()
+	session := s.session
+	s.mu.Unlock()
+	if session == nil {
+		return errors.New("ssh session not started")
+	}
+	return session.WindowChange(rows, cols)
+}
+
+// Reset makes a closed client usable for another Connect: Close leaves the
+// one-shot guard set, and clearing it lets the same client back a reconnect.
+func (s *SSHClient) Reset() {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	s.closed = false
+}
+
+// Close terminates the SSH session and underlying connection. Safe to call
+// multiple times.
+func (s *SSHClient) Close() error {
+	s.mu.Lock()
+	s.gen++
+	if s.dialCancel != nil {
+		s.dialCancel()
+		s.dialCancel = nil
+	}
+	sshClient := s.sshClient
+	session := s.session
+	stdin := s.stdin
+	s.sshClient = nil
+	s.session = nil
+	s.stdin = nil
+	notify := !s.closed
+	s.closed = true
+	listener := s.listener
+	s.mu.Unlock()
+
+	if stdin != nil {
+		if err := stdin.Close(); err != nil {
+			log.Debugf("ssh: stdin close: %v", err)
+		}
+	}
+	if session != nil {
+		if err := session.Close(); err != nil && !errors.Is(err, io.EOF) {
+			log.Debugf("ssh: session close: %v", err)
+		}
+	}
+	var firstErr error
+	if sshClient != nil {
+		if err := sshClient.Close(); err != nil {
+			firstErr = err
+		}
+	}
+	if notify && listener != nil {
+		listener.OnClose("closed by client")
+	}
+	return firstErr
+}
+
+func (s *SSHClient) startSession(cols, rows int) error {
+	log.Debugf("SSH: starting session %dx%d", cols, rows)
+	s.mu.Lock()
+	sshClient := s.sshClient
+	gen := s.gen
+	s.mu.Unlock()
+
+	if sshClient == nil {
+		return errors.New("ssh client not connected")
+	}
+
+	pty, err := nbssh.StartPTYSession(sshClient, cols, rows)
+	if err != nil {
+		return err
+	}
+
+	s.mu.Lock()
+	if gen != s.gen {
+		s.mu.Unlock()
+		closeQuiet(pty.Session, "stale session")
+		return errClientClosed
+	}
+	s.session = pty.Session
+	s.stdin = pty.Stdin
+	s.mu.Unlock()
+
+	readerDone := make(chan string, 2)
+	go func() { readerDone <- s.readLoop(pty.Stdout, "stdout") }()
+	go func() { readerDone <- s.readLoop(pty.Stderr, "stderr") }()
+	go func() {
+		reason := <-readerDone
+		if second := <-readerDone; reason == "" {
+			reason = second
+		}
+		s.notifyClose(gen, reason)
+	}()
+	log.Debug("SSH: session started, shell running")
+	return nil
+}
+
+func (s *SSHClient) buildAuth(cfg *profilemanager.Config, cfgPath string, engine *internal.Engine,
+	serverType detection.ServerType, password string) ([]gossh.AuthMethod, gossh.HostKeyCallback, error) {
+
+	switch serverType {
+	case detection.ServerTypeNetBirdJWT:
+		token, err := s.requestJWTToken(cfg, cfgPath)
+		if err != nil {
+			return nil, nil, fmt.Errorf("jwt: %w", err)
+		}
+		auths := []gossh.AuthMethod{gossh.Password(token)}
+		return auths, nbssh.CreateHostKeyCallback(nbssh.PeerKeyLookup(engine.GetPeerSSHKey)), nil
+
+	case detection.ServerTypeNetBirdNoJWT:
+		if cfg.SSHKey == "" {
+			return nil, nil, errors.New("no NetBird SSH key available")
+		}
+		signer, err := gossh.ParsePrivateKey([]byte(cfg.SSHKey))
+		if err != nil {
+			return nil, nil, fmt.Errorf("parse netbird ssh key: %w", err)
+		}
+		auths := []gossh.AuthMethod{gossh.PublicKeys(signer)}
+		return auths, nbssh.CreateHostKeyCallback(nbssh.PeerKeyLookup(engine.GetPeerSSHKey)), nil
+
+	case detection.ServerTypeRegular:
+		var auths []gossh.AuthMethod
+		if cfg.SSHKey != "" {
+			if signer, err := gossh.ParsePrivateKey([]byte(cfg.SSHKey)); err == nil {
+				auths = append(auths, gossh.PublicKeys(signer))
+			} else {
+				log.Debugf("ssh: parse netbird key for regular auth: %v", err)
+			}
+		}
+		if password != "" {
+			pw := password
+			auths = append(auths, gossh.Password(pw))
+			auths = append(auths, gossh.KeyboardInteractive(func(_, _ string, questions []string, _ []bool) ([]string, error) {
+				answers := make([]string, len(questions))
+				for i := range questions {
+					answers[i] = pw
+				}
+				return answers, nil
+			}))
+		}
+		if len(auths) == 0 {
+			// Nothing to offer at all: ask for a password rather than failing,
+			// so the caller can retry once the user supplies one.
+			return nil, nil, errPasswordRequired
+		}
+		callback, err := s.tofuHostKeyCallback()
+		if err != nil {
+			return nil, nil, err
+		}
+		return auths, callback, nil
+
+	default:
+		return nil, nil, fmt.Errorf("unsupported SSH server type: %v", serverType)
+	}
+}
+
+// tofuHostKeyCallback verifies a regular server's host key against the
+// per-profile known-hosts store. An unknown host returns errHostKeyUnknown so
+// Java can show the fingerprint and, once confirmed, retry with the key
+// trusted; a changed key is rejected outright, as OpenSSH does. When the user
+// has confirmed a fingerprint, the callback accepts exactly that key and
+// appends it to the store.
+func (s *SSHClient) tofuHostKeyCallback() (gossh.HostKeyCallback, error) {
+	s.mu.Lock()
+	configDir := s.knownHostsConfigDir
+	profileID := s.knownHostsProfile
+	trusted := s.trustHostKey
+	s.mu.Unlock()
+
+	if configDir == "" || profileID == "" {
+		return nil, errors.New("no known-hosts store configured for regular SSH")
+	}
+
+	store, err := openKnownHostsStore(configDir, profileID)
+	if err != nil {
+		return nil, fmt.Errorf("load known-hosts store: %w", err)
+	}
+
+	return func(hostname string, remote net.Addr, key gossh.PublicKey) error {
+		verdict, err := store.verify(hostname, remote, key)
+		if err != nil {
+			return err
+		}
+		if verdict == hostKeyMatched {
+			return nil
+		}
+		if verdict == hostKeyChanged {
+			return fmt.Errorf("SSH host key changed for %s (possible attack)", hostname)
+		}
+
+		fingerprint := gossh.FingerprintSHA256(key)
+		if trusted == "" {
+			return &errHostKeyUnknown{fingerprint: fingerprint}
+		}
+		if trusted != fingerprint {
+			return fmt.Errorf("SSH host key changed since it was confirmed for %s", hostname)
+		}
+		if err := store.append(hostname, remote, key); err != nil {
+			return fmt.Errorf("persist trusted host key: %w", err)
+		}
+		// The confirmation is spent: now that the key is stored, a later
+		// reconnect must verify against the file, not re-accept this fingerprint.
+		s.mu.Lock()
+		s.trustHostKey = ""
+		s.mu.Unlock()
+		return nil
+	}, nil
+}
+
+func (s *SSHClient) requestJWTToken(cfg *profilemanager.Config, cfgPath string) (string, error) {
+	s.mu.Lock()
+	urlOpener := s.urlOpener
+	s.mu.Unlock()
+	if urlOpener == nil {
+		return "", errors.New("URL opener not configured for JWT auth")
+	}
+
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Minute)
+	defer cancel()
+
+	flow, err := auth.NewOAuthFlow(ctx, cfg, false, true, profileLoginHint(cfgPath))
+	if err != nil {
+		return "", fmt.Errorf("create oauth flow: %w", err)
+	}
+
+	// The status callback covers the browser round-trip, which would
+	// otherwise leave the terminal blank.
+	tokenInfo, err := runOAuthFlow(ctx, flow, urlOpener, func() {
+		s.notifyStatus("Waiting for browser authentication...")
+	})
+	if err != nil {
+		return "", err
+	}
+
+	token := tokenInfo.GetTokenToUse()
+	if token == "" {
+		return "", errors.New("empty token returned by IdP")
+	}
+
+	// Tells the client the browser round-trip is over so it can dismiss the
+	// surface it opened, the same way the login and session-extend flows do.
+	// Without it the Custom Tab stays in front of the terminal even though the
+	// token has already been collected.
+	urlOpener.OnLoginSuccess()
+
+	return token, nil
+}
+
+func (s *SSHClient) dialAndHandshake(gen uint64, host string, port int, clientConfig *gossh.ClientConfig) error {
+	addr := net.JoinHostPort(host, strconv.Itoa(port))
+	ctx, cancel := context.WithTimeout(context.Background(), sshDialTimeout)
+	defer cancel()
+
+	s.mu.Lock()
+	if gen != s.gen {
+		s.mu.Unlock()
+		return errClientClosed
+	}
+	s.dialCancel = cancel
+	s.mu.Unlock()
+
+	var dialer net.Dialer
+	conn, err := dialer.DialContext(ctx, "tcp", addr)
+	if err != nil {
+		return fmt.Errorf("dial %s: %w", addr, err)
+	}
+
+	client, err := nbssh.Handshake(ctx, conn, addr, clientConfig)
+	if err != nil {
+		return err
+	}
+
+	s.mu.Lock()
+	if gen != s.gen {
+		s.mu.Unlock()
+		closeQuiet(client, "stale ssh client")
+		return errClientClosed
+	}
+	s.sshClient = client
+	listener := s.listener
+	s.mu.Unlock()
+
+	if listener != nil {
+		listener.OnConnected()
+	}
+	return nil
+}
+
+func (s *SSHClient) readLoop(r io.Reader, name string) string {
+	buf := make([]byte, 4096)
+	for {
+		n, err := r.Read(buf)
+		if n > 0 {
+			s.mu.Lock()
+			listener := s.listener
+			s.mu.Unlock()
+			if listener != nil {
+				chunk := make([]byte, n)
+				copy(chunk, buf[:n])
+				listener.OnData(chunk)
+			}
+		}
+		if err != nil {
+			// EOF is a normal shell exit, so report it without a reason.
+			if errors.Is(err, io.EOF) {
+				return ""
+			}
+			log.Debugf("ssh %s read: %v", name, err)
+			return rootCause(err).Error()
+		}
+	}
+}
+
+// notifyStatus writes a progress line to the terminal through the normal
+// output path, so long steps are visible while nothing else is arriving.
+func (s *SSHClient) notifyStatus(text string) {
+	s.mu.Lock()
+	listener := s.listener
+	s.mu.Unlock()
+	if listener != nil {
+		listener.OnData([]byte("\r\n\x1b[33m" + text + "\x1b[0m\r\n"))
+	}
+}
+
+func (s *SSHClient) notifyClose(gen uint64, reason string) {
+	s.mu.Lock()
+	if gen != s.gen || s.closed {
+		s.mu.Unlock()
+		return
+	}
+	s.closed = true
+	listener := s.listener
+	s.mu.Unlock()
+	if listener != nil {
+		listener.OnClose(reason)
+	}
+}
+
+func closeQuiet(c io.Closer, label string) {
+	if c == nil {
+		return
+	}
+	if err := c.Close(); err != nil && !errors.Is(err, io.EOF) {
+		log.Debugf("ssh: close %s: %v", label, err)
+	}
+}
+
+func detectServerType(host string, port int) detection.ServerType {
+	ctx, cancel := context.WithTimeout(context.Background(), sshDetectionTimeout)
+	defer cancel()
+
+	dialer := &net.Dialer{}
+	serverType, err := detection.DetectSSHServerType(ctx, dialer, host, port)
+	if err != nil {
+		log.Debugf("ssh: server detection failed: %v (assuming regular SSH)", err)
+		return detection.ServerTypeRegular
+	}
+	return serverType
+}
+
+// rootCause returns the innermost error of a %w chain, so the terminal shows
+// "i/o timeout" rather than every layer that added context on the way up.
+func rootCause(err error) error {
+	for {
+		// A joined error has no single root, so keep it as-is.
+		if _, ok := err.(interface{ Unwrap() []error }); ok {
+			return err
+		}
+		next := errors.Unwrap(err)
+		if next == nil {
+			return err
+		}
+		err = next
+	}
+}
+
+// isAuthFailure distinguishes credential rejection from dial, timeout and
+// host-key errors, which retrying with a password would not fix.
+func isAuthFailure(err error) bool {
+	if errors.Is(err, errPasswordRequired) {
+		return true
+	}
+	var partial *gossh.PartialSuccessError
+	if errors.As(err, &partial) {
+		return true
+	}
+	return strings.Contains(err.Error(), "unable to authenticate")
+}
+
+// passwordCouldHelp reports whether prompting for a password again can change
+// the outcome. gossh lists a method under "attempted methods" only when the
+// server offered it, so a supplied password that was never attempted means the
+// server does not accept passwords and the real error should surface instead.
+func passwordCouldHelp(err error, passwordOffered bool) bool {
+	if !passwordOffered {
+		return true
+	}
+	msg := err.Error()
+	return strings.Contains(msg, "password") || strings.Contains(msg, "keyboard-interactive")
+}

client/android/ssh_known_hosts.go

@@ -0,0 +1,168 @@
+//go:build android
+
+package android
+
+import (
+	"bytes"
+	"net"
+	"strconv"
+	"strings"
+	"sync"
+
+	gossh "golang.org/x/crypto/ssh"
+	"golang.org/x/crypto/ssh/knownhosts"
+)
+
+const knownHostsNamespace = "ssh"
+
+const (
+	hostKeyUnknown hostKeyVerdict = iota
+	hostKeyMatched
+	hostKeyChanged
+)
+
+var knownHostsMu sync.Mutex
+
+type hostKeyVerdict uint8
+
+type knownHostsSection struct {
+	KnownHosts []string `json:"knownHosts"`
+}
+
+type knownHostsStore struct {
+	prefs prefsStore
+}
+
+// RemoveKnownHost deletes every known-hosts entry for host:port from the
+// profile's store, so a host trusted for a session that is being deleted does
+// not linger. Java calls this only once no session targets that host, so a
+// shared host stays trusted. A missing entry is not an error: the goal state
+// is "absent".
+func RemoveKnownHost(configDir, profileID, host string, port int) error {
+	store, err := openKnownHostsStore(configDir, profileID)
+	if err != nil {
+		return err
+	}
+	return store.removeHost(host, port)
+}
+
+func openKnownHostsStore(configDir, profileID string) (*knownHostsStore, error) {
+	prefs, err := newProfilePrefs(configDir, profileID)
+	if err != nil {
+		return nil, err
+	}
+	return &knownHostsStore{prefs: prefs}, nil
+}
+
+func (st *knownHostsStore) verify(hostname string, remote net.Addr, key gossh.PublicKey) (hostKeyVerdict, error) {
+	lines, err := st.lines()
+	if err != nil {
+		return hostKeyUnknown, err
+	}
+	targets := knownHostsTargets(hostname, remote)
+
+	verdict := hostKeyUnknown
+	for _, line := range lines {
+		pubKey, ok := knownHostsLineKey(line, targets)
+		if !ok {
+			continue
+		}
+		if pubKey.Type() == key.Type() && bytes.Equal(pubKey.Marshal(), key.Marshal()) {
+			return hostKeyMatched, nil
+		}
+		verdict = hostKeyChanged
+	}
+	return verdict, nil
+}
+
+func (st *knownHostsStore) append(hostname string, remote net.Addr, key gossh.PublicKey) error {
+	line := knownhosts.Line(knownHostsTargets(hostname, remote), key)
+
+	knownHostsMu.Lock()
+	defer knownHostsMu.Unlock()
+
+	lines, err := st.lines()
+	if err != nil {
+		return err
+	}
+	return st.prefs.Put(knownHostsNamespace, knownHostsSection{KnownHosts: append(lines, line)})
+}
+
+func (st *knownHostsStore) removeHost(host string, port int) error {
+	target := knownhosts.Normalize(net.JoinHostPort(host, strconv.Itoa(port)))
+
+	knownHostsMu.Lock()
+	defer knownHostsMu.Unlock()
+
+	lines, err := st.lines()
+	if err != nil {
+		return err
+	}
+	kept := make([]string, 0, len(lines))
+	for _, line := range lines {
+		if knownHostsLineMatches(line, target) {
+			continue
+		}
+		kept = append(kept, line)
+	}
+	if len(kept) == len(lines) {
+		return nil
+	}
+	return st.prefs.Put(knownHostsNamespace, knownHostsSection{KnownHosts: kept})
+}
+
+func (st *knownHostsStore) lines() ([]string, error) {
+	var section knownHostsSection
+	if _, err := st.prefs.Get(knownHostsNamespace, &section); err != nil {
+		return nil, err
+	}
+	return section.KnownHosts, nil
+}
+
+func knownHostsTargets(hostname string, remote net.Addr) []string {
+	targets := []string{knownhosts.Normalize(hostname)}
+	if remote != nil {
+		if normalized := knownhosts.Normalize(remote.String()); normalized != targets[0] {
+			targets = append(targets, normalized)
+		}
+	}
+	return targets
+}
+
+func knownHostsLineKey(line string, targets []string) (gossh.PublicKey, bool) {
+	trimmed := strings.TrimSpace(line)
+	if trimmed == "" || strings.HasPrefix(trimmed, "#") {
+		return nil, false
+	}
+	_, hosts, pubKey, _, _, err := gossh.ParseKnownHosts([]byte(trimmed))
+	if err != nil {
+		return nil, false
+	}
+	for _, host := range hosts {
+		for _, target := range targets {
+			if host == target {
+				return pubKey, true
+			}
+		}
+	}
+	return nil, false
+}
+
+// knownHostsLineMatches reports whether a known-hosts line's address list
+// contains the normalized target. Comment and blank lines never match.
+func knownHostsLineMatches(line, target string) bool {
+	trimmed := strings.TrimSpace(line)
+	if trimmed == "" || strings.HasPrefix(trimmed, "#") {
+		return false
+	}
+	fields := strings.Fields(trimmed)
+	if len(fields) == 0 {
+		return false
+	}
+	for _, addr := range strings.Split(fields[0], ",") {
+		if addr == target {
+			return true
+		}
+	}
+	return false
+}

client/android/ssh_sessions.go

@@ -0,0 +1,104 @@
+//go:build android
+
+package android
+
+const (
+	sshSessionsNamespace = "ssh-sessions"
+	maxStoredSSHSessions = 50
+)
+
+type sshSessionRecord struct {
+	ID   string `json:"id"`
+	Host string `json:"host"`
+	Port int    `json:"port"`
+	User string `json:"user"`
+}
+
+type sshSessionsSection struct {
+	Sessions []sshSessionRecord `json:"sessions"`
+}
+
+// SSHSessionEntry is one stored SSH session, without any credential.
+type SSHSessionEntry struct {
+	ID   string
+	Host string
+	Port int
+	User string
+}
+
+// SSHSessionArray wraps stored SSH sessions for gomobile compatibility.
+type SSHSessionArray struct {
+	items []*SSHSessionEntry
+}
+
+// NewSSHSessionArray creates an empty session array to fill via Add.
+func NewSSHSessionArray() *SSHSessionArray {
+	return &SSHSessionArray{}
+}
+
+// Add appends a session entry, oldest first.
+func (a *SSHSessionArray) Add(id, host string, port int, user string) {
+	a.items = append(a.items, &SSHSessionEntry{ID: id, Host: host, Port: port, User: user})
+}
+
+// Length returns the number of entries.
+func (a *SSHSessionArray) Length() int {
+	return len(a.items)
+}
+
+// Get returns the entry at index i, or nil when out of range.
+func (a *SSHSessionArray) Get(i int) *SSHSessionEntry {
+	if i < 0 || i >= len(a.items) {
+		return nil
+	}
+	return a.items[i]
+}
+
+// SSHSessionStore reads and writes a profile's stored SSH sessions.
+type SSHSessionStore struct {
+	prefs prefsStore
+}
+
+// NewSSHSessionStore opens the session store of the given profile.
+func NewSSHSessionStore(configDir, profileID string) (*SSHSessionStore, error) {
+	prefs, err := newProfilePrefs(configDir, profileID)
+	if err != nil {
+		return nil, err
+	}
+	return &SSHSessionStore{prefs: prefs}, nil
+}
+
+// Load returns the stored sessions, oldest first.
+func (s *SSHSessionStore) Load() (*SSHSessionArray, error) {
+	var section sshSessionsSection
+	if _, err := s.prefs.Get(sshSessionsNamespace, &section); err != nil {
+		return nil, err
+	}
+
+	out := NewSSHSessionArray()
+	for _, record := range section.Sessions {
+		if record.ID == "" || record.Host == "" {
+			continue
+		}
+		out.Add(record.ID, record.Host, record.Port, record.User)
+	}
+	return out, nil
+}
+
+// Save replaces the stored sessions, keeping only the newest entries when the
+// list exceeds the storage cap.
+func (s *SSHSessionStore) Save(sessions *SSHSessionArray) error {
+	var items []*SSHSessionEntry
+	if sessions != nil {
+		items = sessions.items
+	}
+	if len(items) > maxStoredSSHSessions {
+		items = items[len(items)-maxStoredSSHSessions:]
+	}
+
+	records := make([]sshSessionRecord, 0, len(items))
+	for _, item := range items {
+		records = append(records, sshSessionRecord{ID: item.ID, Host: item.Host, Port: item.Port, User: item.User})
+	}
+	return s.prefs.Put(sshSessionsNamespace, sshSessionsSection{Sessions: records})
+}

client/anonymize/anonymize.go

@@ -2,6 +2,7 @@ package anonymize
 
 import (
 	"crypto/rand"
+	"encoding/base64"
 	"fmt"
 	"math/big"
 	"net"
@@ -15,13 +16,88 @@ import (
 
 const anonTLD = ".domain"
 
+// Level selects how much the anonymizer redacts. Levels are ordered: a higher
+// level redacts strictly more. On the wire (protos, flags) levels travel as
+// their string form.
+type Level int
+
+const (
+	// LevelDefault anonymizes public IP addresses, IPv6 ULA, domains, and MAC
+	// addresses. Internal IPv4 ranges (RFC 1918, CGNAT, link-local) are
+	// preserved so support can reason about the real topology.
+	LevelDefault Level = iota
+	// LevelStrict additionally anonymizes internal IP ranges, peer names, and
+	// WireGuard public keys.
+	LevelStrict
+)
+
+// LevelDefaultString and LevelStrictString are the wire forms of the levels,
+// for boundaries that pass levels as strings (flags, protos, mobile bindings).
+const (
+	LevelDefaultString = "default"
+	LevelStrictString  = "strict"
+)
+
+// ParseLevel maps s to a Level. Empty means LevelDefault; anything
+// unrecognized maps to LevelStrict so an unknown request never yields less
+// anonymization than intended.
+func ParseLevel(s string) Level {
+	switch strings.ToLower(s) {
+	case "", LevelDefaultString:
+		return LevelDefault
+	default:
+		return LevelStrict
+	}
+}
+
+// String returns the wire form of the level: "default" or "strict".
+func (l Level) String() string {
+	if l >= LevelStrict {
+		return LevelStrictString
+	}
+	return LevelDefaultString
+}
+
+// protectedDomains are NetBird-operated suffixes that stay recognizable in an
+// anonymized bundle. At LevelStrict the labels in front of them (the peer
+// name) are still replaced, except under netbird.io, which only hosts
+// NetBird infrastructure (api, signal, flow), never peer names.
+var protectedDomains = []string{"netbird.io", "netbird.selfhosted", "netbird.cloud", "netbird.stage"}
+
+const infraDomain = "netbird.io"
+
+var (
+	macColonRegex = regexp.MustCompile(`\b[0-9a-fA-F]{2}(?::[0-9a-fA-F]{2}){5}\b`)
+	macDashRegex  = regexp.MustCompile(`\b[0-9a-fA-F]{2}(?:-[0-9a-fA-F]{2}){5}\b`)
+	wgKeyRegex    = regexp.MustCompile(`\b[A-Za-z0-9+/]{43}=`)
+)
+
 type Anonymizer struct {
 	ipAnonymizer     map[netip.Addr]netip.Addr
 	domainAnonymizer map[string]string
-	currentAnonIPv4  netip.Addr
-	currentAnonIPv6  netip.Addr
-	startAnonIPv4    netip.Addr
-	startAnonIPv6    netip.Addr
+	// domainOrder caches the keys of domainAnonymizer sorted longest-first
+	// for AnonymizeString; it is rebuilt when the map gains entries.
+	domainOrder     []string
+	labelAnonymizer map[string]string
+	labelAnonymized map[string]struct{}
+	labelCounter    uint32
+	macAnonymizer   map[string]string
+	macCounter      uint32
+	wgKeyAnonymizer map[string]string
+	wgKeyAnonymized map[string]struct{}
+	currentAnonIPv4 netip.Addr
+	currentAnonIPv6 netip.Addr
+	startAnonIPv4   netip.Addr
+	startAnonIPv6   netip.Addr
+
+	// LevelStrict also anonymizes internal ranges (RFC 1918, CGNAT,
+	// link-local), replacing them from the dedicated internal pools below so
+	// a reader can still tell an internal address from a public one.
+	level                   Level
+	currentAnonInternalIPv4 netip.Addr
+	currentAnonInternalIPv6 netip.Addr
+	startAnonInternalIPv4   netip.Addr
+	startAnonInternalIPv6   netip.Addr
 
 	domainKeyRegex *regexp.Regexp
 }
@@ -32,25 +108,50 @@ func DefaultAddresses() (netip.Addr, netip.Addr) {
 	return netip.AddrFrom4([4]byte{198, 51, 100, 0}), netip.MustParseAddr("2001:db8:ffff::")
 }
 
+// InternalAddresses returns the pool starts used in strict mode for internal
+// ranges. Both are reserved ranges that cannot collide with real addressing:
+// 198.18.0.0 (RFC 2544 benchmarking), 2001:db8:1:: (RFC 3849 documentation).
+func InternalAddresses() (netip.Addr, netip.Addr) {
+	return netip.AddrFrom4([4]byte{198, 18, 0, 0}), netip.MustParseAddr("2001:db8:1::")
+}
+
 func NewAnonymizer(startIPv4, startIPv6 netip.Addr) *Anonymizer {
+	internalIPv4, internalIPv6 := InternalAddresses()
 	return &Anonymizer{
 		ipAnonymizer:     map[netip.Addr]netip.Addr{},
 		domainAnonymizer: map[string]string{},
+		labelAnonymizer:  map[string]string{},
+		labelAnonymized:  map[string]struct{}{},
+		macAnonymizer:    map[string]string{},
+		wgKeyAnonymizer:  map[string]string{},
+		wgKeyAnonymized:  map[string]struct{}{},
 		currentAnonIPv4:  startIPv4,
 		currentAnonIPv6:  startIPv6,
 		startAnonIPv4:    startIPv4,
 		startAnonIPv6:    startIPv6,
 
+		level:                   LevelDefault,
+		currentAnonInternalIPv4: internalIPv4,
+		currentAnonInternalIPv6: internalIPv6,
+		startAnonInternalIPv4:   internalIPv4,
+		startAnonInternalIPv6:   internalIPv6,
+
 		domainKeyRegex: regexp.MustCompile(`\bdomain=([^\s,:"]+)`),
 	}
 }
 
+// SetLevel selects the anonymization level. The zero value of a new
+// Anonymizer is LevelDefault.
+func (a *Anonymizer) SetLevel(level Level) {
+	a.level = level
+}
+
 func (a *Anonymizer) AnonymizeIP(ip netip.Addr) netip.Addr {
+	// Normalize 4-in-6 addresses so ::ffff:192.168.1.1 classifies and maps
+	// like 192.168.1.1.
+	ip = ip.Unmap()
+
 	if ip.IsLoopback() ||
-		ip.IsLinkLocalUnicast() ||
-		ip.IsLinkLocalMulticast() ||
-		ip.IsInterfaceLocalMulticast() ||
-		(ip.Is4() && ip.IsPrivate()) ||
 		ip.IsUnspecified() ||
 		ip.IsMulticast() ||
 		isWellKnown(ip) ||
@@ -59,18 +160,100 @@ func (a *Anonymizer) AnonymizeIP(ip netip.Addr) netip.Addr {
 		return ip
 	}
 
+	if isInternal(ip) && a.level < LevelStrict {
+		return ip
+	}
+
 	if _, ok := a.ipAnonymizer[ip]; !ok {
-		if ip.Is4() {
-			a.ipAnonymizer[ip] = a.currentAnonIPv4
-			a.currentAnonIPv4 = a.currentAnonIPv4.Next()
-		} else {
-			a.ipAnonymizer[ip] = a.currentAnonIPv6
-			a.currentAnonIPv6 = a.currentAnonIPv6.Next()
-		}
+		a.ipAnonymizer[ip] = a.nextAnonIP(ip)
 	}
 	return a.ipAnonymizer[ip]
 }
 
+func (a *Anonymizer) nextAnonIP(ip netip.Addr) netip.Addr {
+	// At the strict level, internal addresses (including IPv6 ULA, matched
+	// by IsPrivate) come from the internal pools so they remain recognizable
+	// as internal without disclosing the real values.
+	if a.level >= LevelStrict && (isInternal(ip) || ip.IsPrivate()) {
+		if ip.Is4() {
+			anon := a.currentAnonInternalIPv4
+			a.currentAnonInternalIPv4 = a.currentAnonInternalIPv4.Next()
+			return anon
+		}
+		anon := a.currentAnonInternalIPv6
+		a.currentAnonInternalIPv6 = a.currentAnonInternalIPv6.Next()
+		return anon
+	}
+
+	if ip.Is4() {
+		anon := a.currentAnonIPv4
+		a.currentAnonIPv4 = a.currentAnonIPv4.Next()
+		return anon
+	}
+	anon := a.currentAnonIPv6
+	a.currentAnonIPv6 = a.currentAnonIPv6.Next()
+	return anon
+}
+
+// AnonymizeMAC replaces a MAC address with a consistent placeholder from the
+// locally administered range starting at 02:00:00:00:00:01, at every
+// anonymization level. Broadcast, multicast, all-zero, and already assigned
+// placeholder addresses are preserved. The colon and dash spellings of the
+// same address share one placeholder; the output keeps the input's separator.
+func (a *Anonymizer) AnonymizeMAC(mac string) string {
+	hw, err := net.ParseMAC(mac)
+	if err != nil || len(hw) != 6 {
+		return mac
+	}
+
+	if isWellKnownMAC(hw) || a.isAnonymizedMAC(hw) {
+		return mac
+	}
+
+	key := hw.String()
+	anon, ok := a.macAnonymizer[key]
+	if !ok {
+		a.macCounter++
+		anon = fmt.Sprintf("02:00:00:%02x:%02x:%02x", byte(a.macCounter>>16), byte(a.macCounter>>8), byte(a.macCounter))
+		a.macAnonymizer[key] = anon
+	}
+
+	if strings.Contains(mac, "-") {
+		anon = strings.ReplaceAll(anon, ":", "-")
+	}
+	return anon
+}
+
+// isAnonymizedMAC reports whether hw is a placeholder this anonymizer already
+// handed out, so a second pass over anonymized output leaves it unchanged.
+func (a *Anonymizer) isAnonymizedMAC(hw net.HardwareAddr) bool {
+	if hw[0] != 0x02 || hw[1] != 0 || hw[2] != 0 {
+		return false
+	}
+	value := uint32(hw[3])<<16 | uint32(hw[4])<<8 | uint32(hw[5])
+	return value <= a.macCounter
+}
+
+// AnonymizeWGKey replaces a WireGuard public key with a consistent random
+// placeholder of the same shape. Keys are only anonymized at LevelStrict;
+// placeholders already handed out pass through unchanged.
+func (a *Anonymizer) AnonymizeWGKey(key string) string {
+	if a.level < LevelStrict || !looksLikeWGKey(key) {
+		return key
+	}
+	if _, ok := a.wgKeyAnonymized[key]; ok {
+		return key
+	}
+
+	anon, ok := a.wgKeyAnonymizer[key]
+	if !ok {
+		anon = generateAnonymousKey()
+		a.wgKeyAnonymizer[key] = anon
+		a.wgKeyAnonymized[anon] = struct{}{}
+	}
+	return anon
+}
+
 func (a *Anonymizer) AnonymizeUDPAddr(addr net.UDPAddr) net.UDPAddr {
 	// Convert IP to netip.Addr
 	ip, ok := netip.AddrFromSlice(addr.IP)
@@ -89,12 +272,12 @@ func (a *Anonymizer) AnonymizeUDPAddr(addr net.UDPAddr) net.UDPAddr {
 
 // isInAnonymizedRange checks if an IP is within the range of already assigned anonymized IPs
 func (a *Anonymizer) isInAnonymizedRange(ip netip.Addr) bool {
-	if ip.Is4() && ip.Compare(a.startAnonIPv4) >= 0 && ip.Compare(a.currentAnonIPv4) <= 0 {
-		return true
-	} else if !ip.Is4() && ip.Compare(a.startAnonIPv6) >= 0 && ip.Compare(a.currentAnonIPv6) <= 0 {
-		return true
+	if ip.Is4() {
+		return inPoolRange(ip, a.startAnonIPv4, a.currentAnonIPv4) ||
+			inPoolRange(ip, a.startAnonInternalIPv4, a.currentAnonInternalIPv4)
 	}
-	return false
+	return inPoolRange(ip, a.startAnonIPv6, a.currentAnonIPv6) ||
+		inPoolRange(ip, a.startAnonInternalIPv6, a.currentAnonInternalIPv6)
 }
 
 func (a *Anonymizer) AnonymizeIPString(ip string) string {
@@ -118,14 +301,23 @@ func (a *Anonymizer) AnonymizeDomain(domain string) string {
 		baseDomain = domain[:len(domain)-1]
 	}
 
-	if strings.HasSuffix(baseDomain, "netbird.io") ||
-		strings.HasSuffix(baseDomain, "netbird.selfhosted") ||
-		strings.HasSuffix(baseDomain, "netbird.cloud") ||
-		strings.HasSuffix(baseDomain, "netbird.stage") ||
-		strings.HasSuffix(baseDomain, anonTLD) {
+	if strings.HasSuffix(baseDomain, anonTLD) {
 		return domain
 	}
 
+	// A reverse zone names an address prefix, so it follows the address rules,
+	// which also keeps its digit labels intact.
+	if zone, ok := a.anonymizeReverseZone(baseDomain); ok {
+		return withTrailingDot(zone, hasDot)
+	}
+
+	if suffix := protectedSuffix(baseDomain); suffix != "" {
+		if a.level < LevelStrict || baseDomain == suffix || suffix == infraDomain {
+			return domain
+		}
+		return withTrailingDot(a.anonymizePeerName(baseDomain, suffix), hasDot)
+	}
+
 	parts := strings.Split(baseDomain, ".")
 	if len(parts) < 2 {
 		return domain
@@ -141,12 +333,53 @@ func (a *Anonymizer) AnonymizeDomain(domain string) string {
 	}
 
 	result := strings.Replace(baseDomain, baseForLookup, anonymized, 1)
-	if hasDot {
-		result += "."
+	if a.level >= LevelStrict && len(parts) > 2 {
+		prefix := strings.TrimSuffix(baseDomain, "."+baseForLookup)
+		result = a.anonymizeLabels(prefix, "host") + "." + anonymized
+		// The full mapping feeds AnonymizeString so seeded FQDNs are caught
+		// in log lines as a whole, labels included.
+		a.domainAnonymizer[baseDomain] = result
+	}
+	return withTrailingDot(result, hasDot)
+}
+
+// anonymizePeerName replaces the labels in front of a protected suffix with
+// numbered peer placeholders, keeping the suffix, and records the full
+// mapping for string replacement in logs. The numbering keeps a peer
+// recognizable across the whole bundle without disclosing its name.
+func (a *Anonymizer) anonymizePeerName(baseDomain, suffix string) string {
+	prefix := strings.TrimSuffix(baseDomain, "."+suffix)
+	result := a.anonymizeLabels(prefix, "peer") + "." + suffix
+	if result != baseDomain {
+		a.domainAnonymizer[baseDomain] = result
 	}
 	return result
 }
 
+// anonymizeLabels replaces each dot-separated label with a consistent
+// numbered placeholder ("<placeholder>-<n>"). Wildcard labels and
+// placeholders already handed out pass through unchanged.
+func (a *Anonymizer) anonymizeLabels(prefix, placeholder string) string {
+	labels := strings.Split(prefix, ".")
+	for i, label := range labels {
+		if label == "*" {
+			continue
+		}
+		if _, ok := a.labelAnonymized[label]; ok {
+			continue
+		}
+		anon, ok := a.labelAnonymizer[label]
+		if !ok {
+			a.labelCounter++
+			anon = fmt.Sprintf("%s-%d", placeholder, a.labelCounter)
+			a.labelAnonymizer[label] = anon
+			a.labelAnonymized[anon] = struct{}{}
+		}
+		labels[i] = anon
+	}
+	return strings.Join(labels, ".")
+}
+
 func (a *Anonymizer) AnonymizeURI(uri string) string {
 	u, err := url.Parse(uri)
 	if err != nil {
@@ -178,17 +411,75 @@ func (a *Anonymizer) AnonymizeString(str string) string {
 	ipv4Regex := regexp.MustCompile(`\b(?:[0-9]{1,3}\.){3}[0-9]{1,3}\b`)
 	ipv6Regex := regexp.MustCompile(`\b([0-9a-fA-F:]+:+[0-9a-fA-F]{0,4})(?:%[0-9a-zA-Z]+)?(?:\/[0-9]{1,3})?(?::[0-9]{1,5})?\b`)
 
+	// Reverse zones go first and are then held out of the passes below: their
+	// labels are digits, which the address patterns would otherwise consume.
+	str, restoreZones := a.replaceReverseZones(str)
+
 	str = ipv4Regex.ReplaceAllStringFunc(str, a.AnonymizeIPString)
 	str = ipv6Regex.ReplaceAllStringFunc(str, a.AnonymizeIPString)
 
-	for domain, anonDomain := range a.domainAnonymizer {
-		str = strings.ReplaceAll(str, domain, anonDomain)
+	for _, domain := range a.sortedDomains() {
+		str = strings.ReplaceAll(str, domain, a.domainAnonymizer[domain])
 	}
 
 	str = a.AnonymizeSchemeURI(str)
 	str = a.AnonymizeDNSLogLine(str)
 
-	return str
+	// MAC handling runs after the IP passes so preserved IPv6 addresses are
+	// already out of the way; the separator guard skips matches embedded in a
+	// longer colon- or dash-separated sequence (such as an IPv6 tail).
+	str = a.anonymizeMACsInString(str, macColonRegex, ':')
+	str = a.anonymizeMACsInString(str, macDashRegex, '-')
+
+	if a.level >= LevelStrict {
+		str = wgKeyRegex.ReplaceAllStringFunc(str, a.AnonymizeWGKey)
+	}
+
+	return restoreZones(str)
+}
+
+// sortedDomains returns the domain mappings longest-first, so a full-FQDN
+// mapping (strict level) is applied before the base-domain mapping it
+// contains. The order is rebuilt only when domainAnonymizer has grown.
+func (a *Anonymizer) sortedDomains() []string {
+	if len(a.domainOrder) == len(a.domainAnonymizer) {
+		return a.domainOrder
+	}
+
+	a.domainOrder = a.domainOrder[:0]
+	for domain := range a.domainAnonymizer {
+		a.domainOrder = append(a.domainOrder, domain)
+	}
+	slices.SortFunc(a.domainOrder, func(x, y string) int {
+		if d := len(y) - len(x); d != 0 {
+			return d
+		}
+		return strings.Compare(x, y)
+	})
+	return a.domainOrder
+}
+
+// anonymizeMACsInString replaces MAC addresses matched by re, skipping
+// matches that directly adjoin another sep so a six-group run inside a longer
+// separated sequence is left alone.
+func (a *Anonymizer) anonymizeMACsInString(str string, re *regexp.Regexp, sep byte) string {
+	matches := re.FindAllStringIndex(str, -1)
+	if len(matches) == 0 {
+		return str
+	}
+
+	var b strings.Builder
+	last := 0
+	for _, m := range matches {
+		if (m[0] > 0 && str[m[0]-1] == sep) || (m[1] < len(str) && str[m[1]] == sep) {
+			continue
+		}
+		b.WriteString(str[last:m[0]])
+		b.WriteString(a.AnonymizeMAC(str[m[0]:m[1]]))
+		last = m[1]
+	}
+	b.WriteString(str[last:])
+	return b.String()
 }
 
 // AnonymizeSchemeURI finds and anonymizes URIs with ws, wss, rel, rels, stun, stuns, turn, and turns schemes.
@@ -239,10 +530,79 @@ func isWellKnown(addr netip.Addr) bool {
 		"128.0.0.0", "8000::", // 2nd split subnet for default routes
 	}
 
-	if slices.Contains(wellKnown, addr.String()) {
+	return slices.Contains(wellKnown, addr.String())
+}
+
+// isInternal reports whether ip identifies a host only within the local
+// network: IPv4 private (RFC 1918), CGNAT (RFC 6598), and link-local (v4 and
+// v6). These are preserved at the default level so support can reason about
+// the real topology, and replaced from the internal pools at the strict
+// level. IPv6 ULA is deliberately not internal: its random global ID uniquely
+// fingerprints the network, so it is anonymized at every level.
+func isInternal(ip netip.Addr) bool {
+	return (ip.Is4() && ip.IsPrivate()) ||
+		ip.IsLinkLocalUnicast() ||
+		isCGNAT(ip)
+}
+
+func inPoolRange(ip, start, current netip.Addr) bool {
+	return ip.Compare(start) >= 0 && ip.Compare(current) <= 0
+}
+
+// isWellKnownMAC reports whether hw carries no stable host identity: all-zero
+// or a group address (broadcast and multicast).
+func isWellKnownMAC(hw net.HardwareAddr) bool {
+	if hw[0]&1 == 1 {
 		return true
 	}
+	for _, b := range hw {
+		if b != 0 {
+			return false
+		}
+	}
+	return true
+}
 
+// looksLikeWGKey reports whether s has the shape of a WireGuard key:
+// 44 base64 characters decoding to 32 bytes.
+func looksLikeWGKey(s string) bool {
+	if len(s) != 44 || s[43] != '=' {
+		return false
+	}
+	decoded, err := base64.StdEncoding.DecodeString(s)
+	return err == nil && len(decoded) == 32
+}
+
+func generateAnonymousKey() string {
+	buf := make([]byte, 32)
+	if _, err := rand.Read(buf); err != nil {
+		return strings.Repeat("A", 43) + "="
+	}
+	return base64.StdEncoding.EncodeToString(buf)
+}
+
+// protectedSuffix returns the protected NetBird suffix baseDomain ends with,
+// or empty. The match is label-anchored so an unrelated domain that merely
+// ends in the same characters is not preserved.
+func protectedSuffix(baseDomain string) string {
+	for _, d := range protectedDomains {
+		if baseDomain == d || strings.HasSuffix(baseDomain, "."+d) {
+			return d
+		}
+	}
+	return ""
+}
+
+func withTrailingDot(domain string, hasDot bool) string {
+	if hasDot {
+		return domain + "."
+	}
+	return domain
+}
+
+// isCGNAT reports whether addr is in 100.64.0.0/10 (RFC 6598), the range
+// NetBird assigns overlay peer addresses from.
+func isCGNAT(addr netip.Addr) bool {
 	cgnatRangeStart := netip.AddrFrom4([4]byte{100, 64, 0, 0})
 	cgnatRange := netip.PrefixFrom(cgnatRangeStart, 10)
 

client/anonymize/anonymize_test.go

@@ -1,8 +1,11 @@
 package anonymize_test
 
 import (
+	"bytes"
+	"encoding/base64"
 	"net/netip"
 	"regexp"
+	"strings"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
@@ -44,6 +47,301 @@ func TestAnonymizeIP(t *testing.T) {
 	}
 }
 
+func TestParseLevel(t *testing.T) {
+	tests := []struct {
+		input  string
+		expect anonymize.Level
+	}{
+		{"", anonymize.LevelDefault},
+		{"default", anonymize.LevelDefault},
+		{"DEFAULT", anonymize.LevelDefault},
+		{"strict", anonymize.LevelStrict},
+		{"STRICT", anonymize.LevelStrict},
+		// Unknown values must never yield less anonymization than requested.
+		{"garbage", anonymize.LevelStrict},
+	}
+
+	for _, tc := range tests {
+		t.Run("input="+tc.input, func(t *testing.T) {
+			assert.Equal(t, tc.expect, anonymize.ParseLevel(tc.input), "parsed level should match")
+		})
+	}
+}
+
+func TestAnonymizeIP_DefaultLevelInternalRanges(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+
+	tests := []struct {
+		name   string
+		ip     string
+		expect string
+	}{
+		{"RFC1918 10/8", "10.1.2.3", "10.1.2.3"},
+		{"RFC1918 172.16/12", "172.16.5.5", "172.16.5.5"},
+		{"RFC1918 192.168/16", "192.168.1.1", "192.168.1.1"},
+		{"CGNAT", "100.64.0.5", "100.64.0.5"},
+		{"IPv4 link-local", "169.254.1.1", "169.254.1.1"},
+		{"IPv6 link-local", "fe80::1", "fe80::1"},
+		// ULA is anonymized even at the default level: its random global ID
+		// uniquely fingerprints the network, unlike shared RFC 1918 space.
+		{"IPv6 ULA", "fd12:3456:789a::1", "2001:db8:ffff::"},
+		// 4-in-6 addresses classify like their unmapped IPv4 form.
+		{"4-in-6 RFC1918", "::ffff:192.168.1.1", "192.168.1.1"},
+		{"4-in-6 CGNAT", "::ffff:100.64.0.5", "100.64.0.5"},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			result := anonymizer.AnonymizeIP(netip.MustParseAddr(tc.ip))
+			assert.Equal(t, tc.expect, result.String(), "default level should preserve internal ranges except ULA")
+		})
+	}
+}
+
+func TestAnonymizeIP_StrictLevel(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+	anonymizer.SetLevel(anonymize.LevelStrict)
+
+	// Order matters: internal pool addresses are assigned sequentially.
+	tests := []struct {
+		name   string
+		ip     string
+		expect string
+	}{
+		{"RFC1918 192.168/16", "192.168.1.1", "198.18.0.0"},
+		{"Second RFC1918", "192.168.1.2", "198.18.0.1"},
+		{"Repeated RFC1918", "192.168.1.1", "198.18.0.0"},
+		{"RFC1918 10/8", "10.1.2.3", "198.18.0.2"},
+		{"RFC1918 172.16/12", "172.16.5.5", "198.18.0.3"},
+		{"CGNAT", "100.64.0.5", "198.18.0.4"},
+		{"IPv4 link-local", "169.254.1.1", "198.18.0.5"},
+		{"Public IPv4 uses public pool", "1.2.3.4", "198.51.100.0"},
+		{"IPv6 link-local", "fe80::1", "2001:db8:1::"},
+		{"IPv6 ULA", "fd12:3456:789a::1", "2001:db8:1::1"},
+		{"Public IPv6 uses public pool", "2607:f8b0:4005:805::200e", "2001:db8:ffff::"},
+		{"Loopback IPv4", "127.0.0.1", "127.0.0.1"},
+		{"Loopback IPv6", "::1", "::1"},
+		{"Unspecified", "0.0.0.0", "0.0.0.0"},
+		{"Multicast", "224.0.0.251", "224.0.0.251"},
+		{"Well known resolver", "8.8.8.8", "8.8.8.8"},
+		{"Well known split marker", "128.0.0.0", "128.0.0.0"},
+		{"In internal pool range", "198.18.0.3", "198.18.0.3"},
+		{"In public pool range", "198.51.100.0", "198.51.100.0"},
+		{"4-in-6 repeated RFC1918", "::ffff:192.168.1.1", "198.18.0.0"},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			result := anonymizer.AnonymizeIP(netip.MustParseAddr(tc.ip))
+			assert.Equal(t, tc.expect, result.String(), "strict level should replace internal ranges from the internal pools")
+		})
+	}
+}
+
+func TestAnonymizeString_StrictInternalIPs(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+	anonymizer.SetLevel(anonymize.LevelStrict)
+
+	input := "route 10.20.30.0/24 via 192.168.1.1 dev eth0 src 100.64.0.7"
+	firstPass := anonymizer.AnonymizeString(input)
+	secondPass := anonymizer.AnonymizeString(firstPass)
+
+	assert.NotContains(t, firstPass, "10.20.30.0", "private network address should be anonymized")
+	assert.NotContains(t, firstPass, "192.168.1.1", "private gateway should be anonymized")
+	assert.NotContains(t, firstPass, "100.64.0.7", "CGNAT address should be anonymized")
+	assert.Contains(t, firstPass, "/24", "prefix length should be preserved")
+	assert.Equal(t, firstPass, secondPass, "second pass should not further anonymize the string")
+}
+
+func TestAnonymizeMAC(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+
+	first := anonymizer.AnonymizeMAC("aa:bb:cc:dd:ee:0f")
+	assert.Equal(t, "02:00:00:00:00:01", first, "first MAC should get the first placeholder")
+	assert.Equal(t, first, anonymizer.AnonymizeMAC("aa:bb:cc:dd:ee:0f"), "repeated MAC should map to the same placeholder")
+	assert.Equal(t, first, anonymizer.AnonymizeMAC("AA:BB:CC:DD:EE:0F"), "case should not affect the mapping")
+	assert.Equal(t, "02-00-00-00-00-01", anonymizer.AnonymizeMAC("AA-BB-CC-DD-EE-0F"), "dash form should keep its separator but share the mapping")
+
+	second := anonymizer.AnonymizeMAC("10:22:33:44:55:66")
+	assert.Equal(t, "02:00:00:00:00:02", second, "second distinct MAC should get the next placeholder")
+
+	tests := []struct {
+		name string
+		mac  string
+	}{
+		{"Broadcast", "ff:ff:ff:ff:ff:ff"},
+		{"IPv4 multicast", "01:00:5e:00:00:fb"},
+		{"IPv6 multicast", "33:33:00:00:00:01"},
+		{"All zero", "00:00:00:00:00:00"},
+		{"Assigned placeholder", "02:00:00:00:00:01"},
+		{"Invalid", "not-a-mac"},
+	}
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			assert.Equal(t, tc.mac, anonymizer.AnonymizeMAC(tc.mac), "should be preserved")
+		})
+	}
+}
+
+func TestAnonymizeString_MACAddresses(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+
+	tests := []struct {
+		name   string
+		input  string
+		expect string
+	}{
+		{
+			name:   "nftables ether rule",
+			input:  "ether saddr aa:bb:cc:dd:ee:ff drop",
+			expect: "ether saddr 02:00:00:00:00:01 drop",
+		},
+		{
+			name:   "Windows dash form",
+			input:  "Physical Address : AA-BB-CC-DD-EE-FF",
+			expect: "Physical Address : 02-00-00-00-00-01",
+		},
+		{
+			name:   "IPv6 address tail is not treated as MAC",
+			input:  "addr fe80:0:11:22:33:44:55:66 scope link",
+			expect: "addr fe80:0:11:22:33:44:55:66 scope link",
+		},
+		{
+			name:   "broadcast MAC preserved",
+			input:  "dst ff:ff:ff:ff:ff:ff type ARP",
+			expect: "dst ff:ff:ff:ff:ff:ff type ARP",
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			result := anonymizer.AnonymizeString(tc.input)
+			assert.Equal(t, tc.expect, result, "MAC addresses should be anonymized at every level")
+			assert.Equal(t, result, anonymizer.AnonymizeString(result), "second pass should not change the result")
+		})
+	}
+}
+
+func TestAnonymizeWGKey(t *testing.T) {
+	key := base64.StdEncoding.EncodeToString(bytes.Repeat([]byte{0x42}, 32))
+
+	t.Run("default level preserves keys", func(t *testing.T) {
+		anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+		assert.Equal(t, key, anonymizer.AnonymizeWGKey(key), "default level should not touch WireGuard keys")
+	})
+
+	t.Run("strict level replaces keys", func(t *testing.T) {
+		anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+		anonymizer.SetLevel(anonymize.LevelStrict)
+
+		anon := anonymizer.AnonymizeWGKey(key)
+		assert.NotEqual(t, key, anon, "strict level should replace the key")
+		assert.Regexp(t, `^[A-Za-z0-9+/]{43}=$`, anon, "placeholder should keep the WireGuard key shape")
+		assert.Equal(t, anon, anonymizer.AnonymizeWGKey(key), "repeated key should map to the same placeholder")
+		assert.Equal(t, anon, anonymizer.AnonymizeWGKey(anon), "an assigned placeholder should pass through unchanged")
+
+		assert.Equal(t, "not-a-key", anonymizer.AnonymizeWGKey("not-a-key"), "non-key values should be preserved")
+	})
+}
+
+func TestAnonymizeString_WGKeys(t *testing.T) {
+	key := base64.StdEncoding.EncodeToString(bytes.Repeat([]byte{0x42}, 32))
+	input := "peer " + key + " handshake completed"
+
+	t.Run("default level preserves keys", func(t *testing.T) {
+		anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+		assert.Equal(t, input, anonymizer.AnonymizeString(input), "default level should not touch WireGuard keys in strings")
+	})
+
+	t.Run("strict level replaces keys", func(t *testing.T) {
+		anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+		anonymizer.SetLevel(anonymize.LevelStrict)
+
+		firstPass := anonymizer.AnonymizeString(input)
+		assert.NotContains(t, firstPass, key, "the key should not survive strict anonymization")
+		assert.Equal(t, anonymizer.AnonymizeWGKey(key), extractKey(t, firstPass), "string replacement should be consistent with AnonymizeWGKey")
+		assert.Equal(t, firstPass, anonymizer.AnonymizeString(firstPass), "second pass should not change the result")
+	})
+}
+
+func extractKey(t *testing.T, logLine string) string {
+	t.Helper()
+	fields := strings.Fields(logLine)
+	require.Len(t, fields, 4, "log line should keep its structure")
+	return fields[1]
+}
+
+func TestAnonymizeDomain_StrictLevel(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+	anonymizer.SetLevel(anonymize.LevelStrict)
+
+	t.Run("netbird peer name", func(t *testing.T) {
+		result := anonymizer.AnonymizeDomain("my-laptop.netbird.cloud")
+		assert.Regexp(t, `^peer-\d+\.netbird\.cloud$`, result, "peer name should be anonymized, suffix kept")
+		assert.NotContains(t, result, "my-laptop", "the peer name should not survive")
+		assert.Equal(t, result, anonymizer.AnonymizeDomain("my-laptop.netbird.cloud"), "repeated domain should map consistently")
+		assert.Equal(t, result, anonymizer.AnonymizeDomain(result), "an anonymized domain should pass through unchanged")
+	})
+
+	t.Run("bare netbird domain", func(t *testing.T) {
+		assert.Equal(t, "netbird.cloud", anonymizer.AnonymizeDomain("netbird.cloud"), "the bare protected suffix should be preserved")
+	})
+
+	t.Run("netbird infrastructure preserved", func(t *testing.T) {
+		assert.Equal(t, "api.netbird.io", anonymizer.AnonymizeDomain("api.netbird.io"),
+			"netbird.io hosts infrastructure, not peer names, and should stay readable")
+	})
+
+	t.Run("leading labels of other domains", func(t *testing.T) {
+		result := anonymizer.AnonymizeDomain("host1.corp.example.com")
+		assert.Regexp(t, `^host-\d+\.host-\d+\.anon-[a-zA-Z0-9]+\.domain$`, result, "every label should be anonymized")
+		for _, label := range []string{"host1", "corp", "example"} {
+			assert.NotContains(t, result, label, "no original label should survive")
+		}
+		assert.Equal(t, result, anonymizer.AnonymizeDomain("host1.corp.example.com"), "repeated domain should map consistently")
+	})
+
+	t.Run("same label maps consistently across domains", func(t *testing.T) {
+		first := anonymizer.AnonymizeDomain("shared.one.com")
+		second := anonymizer.AnonymizeDomain("shared.two.com")
+		assert.Equal(t, strings.Split(first, ".")[0], strings.Split(second, ".")[0], "the shared host label should get one placeholder")
+	})
+
+	t.Run("wildcard label preserved", func(t *testing.T) {
+		result := anonymizer.AnonymizeDomain("*.example.com")
+		assert.Regexp(t, `^\*\.anon-[a-zA-Z0-9]+\.domain$`, result, "the wildcard label should stay a wildcard")
+	})
+}
+
+func TestAnonymizeDomain_DefaultLevelKeepsPeerNames(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+
+	assert.Equal(t, "my-laptop.netbird.cloud", anonymizer.AnonymizeDomain("my-laptop.netbird.cloud"),
+		"default level should preserve netbird FQDNs including the peer name")
+	assert.Regexp(t, `^sub\.anon-[a-zA-Z0-9]+\.domain$`, anonymizer.AnonymizeDomain("sub.example.com"),
+		"default level should keep subdomain labels")
+}
+
+func TestAnonymizeString_StrictPeerNames(t *testing.T) {
+	anonymizer := anonymize.NewAnonymizer(anonymize.DefaultAddresses())
+	anonymizer.SetLevel(anonymize.LevelStrict)
+
+	// Seed like the bundle generator does from the status: base first, then
+	// the full FQDN, so replacement must prefer the longer mapping.
+	anonBase := anonymizer.AnonymizeDomain("example.com")
+	anonPeer := anonymizer.AnonymizeDomain("peer1.netbird.cloud")
+	anonHost := anonymizer.AnonymizeDomain("host1.example.com")
+
+	logLine := "connected to peer1.netbird.cloud via host1.example.com endpoint"
+	firstPass := anonymizer.AnonymizeString(logLine)
+	assert.NotContains(t, firstPass, "peer1", "the peer name should not survive in logs")
+	assert.NotContains(t, firstPass, "host1", "the host label should not survive in logs")
+	assert.Contains(t, firstPass, anonPeer, "the seeded peer mapping should be applied")
+	assert.Contains(t, firstPass, anonHost, "the seeded host mapping should be applied, not just the base mapping")
+	assert.NotContains(t, firstPass, "host1."+anonBase, "the base mapping must not preempt the longer FQDN mapping")
+	assert.Equal(t, firstPass, anonymizer.AnonymizeString(firstPass), "second pass should not change the result")
+}
+
 func TestAnonymizeDNSLogLine(t *testing.T) {
 	anonymizer := anonymize.NewAnonymizer(netip.Addr{}, netip.Addr{})
 	tests := []struct {

client/anonymize/reverse_zone.go

@@ -0,0 +1,174 @@
+package anonymize
+
+import (
+	"encoding/hex"
+	"net/netip"
+	"regexp"
+	"strconv"
+	"strings"
+)
+
+const (
+	reverseZoneSuffixV4 = ".in-addr.arpa"
+	reverseZoneSuffixV6 = ".ip6.arpa"
+
+	v6Nibbles = 32
+	v4Octets  = 4
+)
+
+// reverseZoneRegexes match a reverse zone or a full reverse name in free text.
+// They are applied before the address passes of AnonymizeString, whose IPv4
+// pattern would otherwise consume the digit labels of a zone and replace parts
+// of it with unrelated addresses.
+var reverseZoneRegexes = []*regexp.Regexp{
+	regexp.MustCompile(`(?:[0-9]{1,3}\.){1,4}in-addr\.arpa\b`),
+	regexp.MustCompile(`(?:[0-9a-fA-F]\.){1,32}ip6\.arpa\b`),
+}
+
+// anonymizeReverseZone maps a reverse zone to the zone of the anonymized form
+// of the prefix it encodes, so it follows the address rules rather than the
+// domain ones: the zone of an address that is preserved is preserved too, and
+// the zone of one that is replaced names the replacement. This keeps a reverse
+// zone recognizable as such, and consistent with the addresses it belongs to
+// elsewhere in the same output. It reports false for anything that is not a
+// reverse zone.
+func (a *Anonymizer) anonymizeReverseZone(domain string) (string, bool) {
+	prefix, labelCount, suffix, ok := parseReverseZone(domain)
+	if !ok {
+		return "", false
+	}
+
+	anonymized := a.AnonymizeIP(prefix)
+	if anonymized == prefix {
+		return domain, true
+	}
+
+	return reverseZoneName(anonymized, labelCount) + suffix, true
+}
+
+// replaceReverseZones anonymizes every reverse zone in str and swaps each one
+// for a placeholder, returning a function that puts the anonymized zones back.
+// The placeholders carry no dots, digits or colons, so no later pass matches
+// them.
+func (a *Anonymizer) replaceReverseZones(str string) (string, func(string) string) {
+	var zones []string
+
+	for _, re := range reverseZoneRegexes {
+		str = re.ReplaceAllStringFunc(str, func(match string) string {
+			zone, ok := a.anonymizeReverseZone(match)
+			if !ok {
+				return match
+			}
+
+			zones = append(zones, zone)
+			return reverseZonePlaceholder(len(zones) - 1)
+		})
+	}
+
+	if len(zones) == 0 {
+		return str, func(s string) string { return s }
+	}
+
+	return str, func(s string) string {
+		for i, zone := range zones {
+			s = strings.ReplaceAll(s, reverseZonePlaceholder(i), zone)
+		}
+		return s
+	}
+}
+
+func reverseZonePlaceholder(index int) string {
+	return "\x00reversezone" + strconv.Itoa(index) + "\x00"
+}
+
+// parseReverseZone turns a reverse zone into the address of the prefix its
+// labels spell backwards, padding the absent low-order part with zeroes, and
+// returns the label count and zone suffix so the name can be rebuilt.
+func parseReverseZone(domain string) (netip.Addr, int, string, bool) {
+	lower := strings.ToLower(domain)
+
+	switch {
+	case strings.HasSuffix(lower, reverseZoneSuffixV4):
+		labels := strings.Split(strings.TrimSuffix(lower, reverseZoneSuffixV4), ".")
+		addr, ok := reverseZoneAddrV4(labels)
+		return addr, len(labels), reverseZoneSuffixV4, ok
+	case strings.HasSuffix(lower, reverseZoneSuffixV6):
+		labels := strings.Split(strings.TrimSuffix(lower, reverseZoneSuffixV6), ".")
+		addr, ok := reverseZoneAddrV6(labels)
+		return addr, len(labels), reverseZoneSuffixV6, ok
+	default:
+		return netip.Addr{}, 0, "", false
+	}
+}
+
+func reverseZoneAddrV4(labels []string) (netip.Addr, bool) {
+	if len(labels) == 0 || len(labels) > v4Octets {
+		return netip.Addr{}, false
+	}
+
+	var octets [v4Octets]byte
+	for i, label := range labels {
+		octet, err := strconv.ParseUint(label, 10, 8)
+		if err != nil {
+			return netip.Addr{}, false
+		}
+		octets[len(labels)-1-i] = byte(octet)
+	}
+
+	return netip.AddrFrom4(octets), true
+}
+
+func reverseZoneAddrV6(labels []string) (netip.Addr, bool) {
+	if len(labels) == 0 || len(labels) > v6Nibbles {
+		return netip.Addr{}, false
+	}
+
+	nibbles := make([]byte, 0, v6Nibbles)
+	for i := len(labels) - 1; i >= 0; i-- {
+		if len(labels[i]) != 1 || !isHexDigit(labels[i][0]) {
+			return netip.Addr{}, false
+		}
+		nibbles = append(nibbles, labels[i][0])
+	}
+	for len(nibbles) < v6Nibbles {
+		nibbles = append(nibbles, '0')
+	}
+
+	var groups []string
+	for i := 0; i < len(nibbles); i += 4 {
+		groups = append(groups, string(nibbles[i:i+4]))
+	}
+
+	addr, err := netip.ParseAddr(strings.Join(groups, ":"))
+	if err != nil {
+		return netip.Addr{}, false
+	}
+
+	return addr, true
+}
+
+// reverseZoneName spells the first labelCount labels of addr backwards, the
+// inverse of parseReverseZone, without the zone suffix.
+func reverseZoneName(addr netip.Addr, labelCount int) string {
+	labels := make([]string, 0, labelCount)
+
+	if addr.Is4() {
+		octets := addr.As4()
+		for i := labelCount - 1; i >= 0; i-- {
+			labels = append(labels, strconv.Itoa(int(octets[i])))
+		}
+		return strings.Join(labels, ".")
+	}
+
+	address := addr.As16()
+	nibbles := hex.EncodeToString(address[:])
+	for i := labelCount - 1; i >= 0; i-- {
+		labels = append(labels, string(nibbles[i]))
+	}
+
+	return strings.Join(labels, ".")
+}
+
+func isHexDigit(c byte) bool {
+	return c >= '0' && c <= '9' || c >= 'a' && c <= 'f' || c >= 'A' && c <= 'F'
+}

client/anonymize/reverse_zone_test.go

@@ -0,0 +1,171 @@
+package anonymize
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func newLeveledAnonymizer(level Level) *Anonymizer {
+	a := NewAnonymizer(DefaultAddresses())
+	a.SetLevel(level)
+	return a
+}
+
+// TestAnonymizeDomainReverseZone covers reverse zones going through the address
+// rules instead of the domain ones, so a zone stays a zone and an address that
+// is preserved keeps the zone that names it.
+func TestAnonymizeDomainReverseZone(t *testing.T) {
+	// 100.64.0.0/10 is the overlay range, which is CGNAT: preserved at the
+	// default level and replaced from the internal pool at the strict one
+	const overlayZone = "64.100.in-addr.arpa"
+
+	t.Run("overlay zone preserved at the default level", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+		assert.Equal(t, overlayZone, a.AnonymizeDomain(overlayZone), "should keep the zone of a preserved address")
+	})
+
+	t.Run("private zone preserved at the default level", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+		assert.Equal(t, "168.192.in-addr.arpa", a.AnonymizeDomain("168.192.in-addr.arpa"), "should keep the zone of a private address")
+	})
+
+	t.Run("overlay zone replaced at the strict level", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelStrict)
+
+		got := a.AnonymizeDomain(overlayZone)
+		require.True(t, strings.HasSuffix(got, reverseZoneSuffixV4), "should stay a reverse zone, got %q", got)
+		assert.NotEqual(t, overlayZone, got, "should replace the encoded prefix")
+		assert.Len(t, strings.Split(strings.TrimSuffix(got, reverseZoneSuffixV4), "."), 2,
+			"should keep the label count, got %q", got)
+	})
+
+	t.Run("public zone replaced at the default level", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		got := a.AnonymizeDomain("113.0.203.in-addr.arpa")
+		require.True(t, strings.HasSuffix(got, reverseZoneSuffixV4), "should stay a reverse zone, got %q", got)
+		assert.NotEqual(t, "113.0.203.in-addr.arpa", got, "should replace a public prefix")
+	})
+
+	t.Run("zone of an address keeps that address mapping", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		anonymizedAddr := a.AnonymizeIPString("203.0.113.7")
+		got := a.AnonymizeDomain("7.113.0.203.in-addr.arpa")
+
+		octets := strings.Split(anonymizedAddr, ".")
+		want := octets[3] + "." + octets[2] + "." + octets[1] + "." + octets[0] + reverseZoneSuffixV4
+		assert.Equal(t, want, got, "should name the same replacement as the address itself")
+	})
+
+	t.Run("ipv6 nibble labels stay single digits", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		zone := "0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0" + reverseZoneSuffixV6
+		got := a.AnonymizeDomain(zone)
+
+		require.True(t, strings.HasSuffix(got, reverseZoneSuffixV6), "should stay a reverse zone, got %q", got)
+		labels := strings.Split(strings.TrimSuffix(got, reverseZoneSuffixV6), ".")
+		assert.Len(t, labels, 28, "should keep every nibble label, got %q", got)
+		for _, label := range labels {
+			assert.Len(t, label, 1, "nibble label %q should stay a single digit", label)
+		}
+	})
+
+	t.Run("trailing dot is kept", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+		assert.Equal(t, "64.100.in-addr.arpa.", a.AnonymizeDomain("64.100.in-addr.arpa."), "should keep the trailing dot")
+	})
+
+	t.Run("a domain that only looks like a zone is anonymized as a domain", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		got := a.AnonymizeDomain("not-a-zone.in-addr.arpa")
+		assert.NotContains(t, got, "in-addr.arpa", "should fall back to domain anonymization")
+	})
+}
+
+// TestAnonymizeStringReverseZone verifies that a zone inside free text, such as
+// a DNS log line, is not chewed up by the address passes. The IPv4 pattern
+// matches any run of dotted digits, which a reverse zone is made of.
+func TestAnonymizeStringReverseZone(t *testing.T) {
+	t.Run("ipv6 zone survives the address passes", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		zone := "0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0" + reverseZoneSuffixV6
+		got := a.AnonymizeString("question: domain=" + zone + " type=PTR")
+
+		assert.Contains(t, got, "type=PTR", "should keep the rest of the line")
+		assert.NotContains(t, got, "198.51.100", "should not rewrite nibble labels as an address")
+
+		labels := strings.Split(strings.TrimSuffix(strings.TrimPrefix(got, "question: domain="), reverseZoneSuffixV6+" type=PTR"), ".")
+		assert.Len(t, labels, 28, "should keep every nibble label, got %q", got)
+	})
+
+	t.Run("preserved ipv4 zone is untouched", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		line := "reverse zone 64.100.in-addr.arpa registered"
+		assert.Equal(t, line, a.AnonymizeString(line), "should keep the zone of a preserved address")
+	})
+
+	t.Run("public ipv4 zone is replaced consistently", func(t *testing.T) {
+		a := newLeveledAnonymizer(LevelDefault)
+
+		got := a.AnonymizeString("zone 113.0.203.in-addr.arpa and address 203.0.113.7")
+		assert.NotContains(t, got, "113.0.203.in-addr.arpa", "should replace the zone")
+		assert.NotContains(t, got, "203.0.113.7", "should replace the address")
+		assert.Contains(t, got, reverseZoneSuffixV4, "should keep the zone suffix")
+	})
+}
+
+func TestParseReverseZone(t *testing.T) {
+	tests := []struct {
+		name   string
+		zone   string
+		addr   string
+		labels int
+	}{
+		{name: "v4 two labels", zone: "0.100" + reverseZoneSuffixV4, addr: "100.0.0.0", labels: 2},
+		{name: "v4 three labels", zone: "1.168.192" + reverseZoneSuffixV4, addr: "192.168.1.0", labels: 3},
+		{name: "v4 full address", zone: "7.113.0.203" + reverseZoneSuffixV4, addr: "203.0.113.7", labels: 4},
+		{
+			name:   "v6 prefix",
+			zone:   "0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.2.0.0.0" + reverseZoneSuffixV6,
+			addr:   "2::",
+			labels: 28,
+		},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			addr, labels, suffix, ok := parseReverseZone(tc.zone)
+			require.True(t, ok, "should decode the reverse zone")
+			assert.Equal(t, tc.addr, addr.String(), "should decode to the encoded prefix")
+			assert.Equal(t, tc.labels, labels, "should count the labels")
+			assert.Equal(t, tc.zone, reverseZoneName(addr, labels)+suffix, "should re-encode to the original zone")
+		})
+	}
+}
+
+func TestParseReverseZoneRejectsNonZones(t *testing.T) {
+	tests := []string{
+		"example.com",
+		"in-addr.arpa",
+		"x.100" + reverseZoneSuffixV4,
+		"256" + reverseZoneSuffixV4,
+		"1.2.3.4.5" + reverseZoneSuffixV4,
+		"ab" + reverseZoneSuffixV6,
+		"g" + reverseZoneSuffixV6,
+	}
+
+	for _, zone := range tests {
+		t.Run(zone, func(t *testing.T) {
+			_, _, _, ok := parseReverseZone(zone)
+			assert.False(t, ok, "should reject %q", zone)
+		})
+	}
+}

client/cmd/debug.go

@@ -3,7 +3,6 @@ package cmd
 import (
 	"context"
 	"fmt"
-	"os/user"
 	"strings"
 	"time"
 
@@ -27,8 +26,8 @@ import (
 const errCloseConnection = "Failed to close connection: %v"
 
 var (
-	logFileCount        uint32
-	systemInfoFlag      bool
+	logFileCount             uint32
+	systemInfoFlag           bool
 	uploadBundleFlag         bool
 	uploadBundleURLFlag      string
 	uploadBundleInsecureFlag bool
@@ -114,7 +113,7 @@ func debugConfigDump(cmd *cobra.Command, _ []string) error {
 	if err != nil {
 		return fmt.Errorf("get active profile: %v", err)
 	}
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %v", err)
 	}
@@ -156,6 +155,11 @@ func debugConfigDump(cmd *cobra.Command, _ []string) error {
 // request. Returns an error if the RPC fails or if the daemon reports
 // an upload failure reason.
 func debugBundle(cmd *cobra.Command, _ []string) error {
+	anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
+	if err != nil {
+		return err
+	}
+
 	conn, err := getClient(cmd)
 	if err != nil {
 		return err
@@ -168,10 +172,11 @@ func debugBundle(cmd *cobra.Command, _ []string) error {
 
 	client := proto.NewDaemonServiceClient(conn)
 	request := &proto.DebugBundleRequest{
-		Anonymize:    anonymizeFlag,
-		SystemInfo:   systemInfoFlag,
-		LogFileCount: logFileCount,
-		CliVersion:   version.NetbirdVersion(),
+		Anonymize:      anonymizeEnabled,
+		AnonymizeLevel: anonymizeLevel.String(),
+		SystemInfo:     systemInfoFlag,
+		LogFileCount:   logFileCount,
+		CliVersion:     version.NetbirdVersion(),
 	}
 	if uploadBundleFlag {
 		request.UploadURL = uploadBundleURLFlag
@@ -229,6 +234,11 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 		return fmt.Errorf("invalid duration format: %v", err)
 	}
 
+	anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
+	if err != nil {
+		return err
+	}
+
 	conn, err := getClient(cmd)
 	if err != nil {
 		return err
@@ -368,10 +378,11 @@ func runForDuration(cmd *cobra.Command, args []string) error {
 	cmd.Println("Creating debug bundle...")
 
 	request := &proto.DebugBundleRequest{
-		Anonymize:    anonymizeFlag,
-		SystemInfo:   systemInfoFlag,
-		LogFileCount: logFileCount,
-		CliVersion:   version.NetbirdVersion(),
+		Anonymize:      anonymizeEnabled,
+		AnonymizeLevel: anonymizeLevel.String(),
+		SystemInfo:     systemInfoFlag,
+		LogFileCount:   logFileCount,
+		CliVersion:     version.NetbirdVersion(),
 	}
 	if uploadBundleFlag {
 		request.UploadURL = uploadBundleURLFlag

client/cmd/jobs.go

@@ -0,0 +1,13 @@
+package cmd
+
+// remoteJobsAllowedFlag opts this peer into running remote jobs (e.g. debug
+// bundles) requested by the management server. It defaults to false: remote
+// jobs are an explicit opt-in, and enabling it is a privileged change (see the
+// daemon gate in client/server), mirroring the SSH server opt-in.
+const remoteJobsAllowedFlag = "allow-remote-jobs"
+
+var remoteJobsAllowed bool
+
+func init() {
+	upCmd.PersistentFlags().BoolVar(&remoteJobsAllowed, remoteJobsAllowedFlag, false, "Allow the management server to run remote jobs (e.g. debug bundles) on this peer")
+}

client/cmd/login.go

@@ -4,8 +4,6 @@ import (
 	"context"
 	"fmt"
 	"os"
-	"os/user"
-	"runtime"
 	"strings"
 
 	log "github.com/sirupsen/logrus"
@@ -54,7 +52,7 @@ var loginCmd = &cobra.Command{
 			// nolint
 			ctx = context.WithValue(ctx, system.DeviceNameCtxKey, hostName)
 		}
-		username, err := user.Current()
+		username, err := profilemanager.InvokingUser()
 		if err != nil {
 			return fmt.Errorf("get current user: %v", err)
 		}
@@ -75,7 +73,7 @@ var loginCmd = &cobra.Command{
 			if providedSetupKey != "" {
 				return fmt.Errorf("--extend cannot be combined with a setup key; setup keys can only enrol new peers")
 			}
-			if err := doExtendSession(ctx, cmd); err != nil {
+			if err := doExtendSession(ctx, cmd, activeProf); err != nil {
 				return fmt.Errorf("extend session failed: %v", err)
 			}
 			return nil
@@ -93,7 +91,7 @@ var loginCmd = &cobra.Command{
 			return fmt.Errorf("daemon login failed: %v", err)
 		}
 
-		cmd.Println("Logging successfully")
+		cmd.Println("Login successful")
 
 		return nil
 	},
@@ -121,7 +119,7 @@ func doDaemonLogin(ctx context.Context, cmd *cobra.Command, providedSetupKey str
 	loginRequest := proto.LoginRequest{
 		SetupKey:            providedSetupKey,
 		ManagementUrl:       managementURL,
-		IsUnixDesktopClient: isUnixRunningDesktop(),
+		IsUnixDesktopClient: util.HasGraphicalSession(),
 		Hostname:            hostName,
 		DnsLabels:           dnsLabelsReq,
 		ProfileName:         &handle,
@@ -177,7 +175,7 @@ func doDaemonLogin(ctx context.Context, cmd *cobra.Command, providedSetupKey str
 // (browser + verification URL) and the resulting JWT is forwarded to the
 // management server's ExtendAuthSession RPC. The tunnel stays up
 // throughout — no Down/Up, no network-map resync.
-func doExtendSession(ctx context.Context, cmd *cobra.Command) error {
+func doExtendSession(ctx context.Context, cmd *cobra.Command, activeProf *profilemanager.Profile) error {
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
 		//nolint
@@ -189,15 +187,14 @@ func doExtendSession(ctx context.Context, cmd *cobra.Command) error {
 
 	client := proto.NewDaemonServiceClient(conn)
 
-	req := &proto.RequestExtendAuthSessionRequest{}
-	// Pre-fill the IdP login hint from the active profile so the user
+	// the CLI runs in the user's session, the daemon does not: tell it what we can see
+	req := &proto.RequestExtendAuthSessionRequest{HasGraphicalSession: util.HasGraphicalSession()}
+	// Pre-fill the IdP login hint from the resolved profile so the user
 	// doesn't have to retype their email. Best-effort: we still proceed
 	// without a hint if the lookup fails.
 	pm := profilemanager.NewProfileManager()
-	if active, perr := pm.GetActiveProfile(); perr == nil {
-		if profState, sperr := pm.GetProfileState(active.ID); sperr == nil && profState.Email != "" {
-			req.Hint = &profState.Email
-		}
+	if profState, perr := pm.GetProfileState(activeProf.ID); perr == nil && profState.Email != "" {
+		req.Hint = &profState.Email
 	}
 
 	startResp, err := client.RequestExtendAuthSession(ctx, req)
@@ -235,9 +232,11 @@ func getActiveProfile(ctx context.Context, pm *profilemanager.ProfileManager, pr
 	// switch profile if provided
 
 	if profileName != "" {
-		if err := switchProfileOnDaemon(ctx, pm, profileName, username); err != nil {
+		prof, err := switchProfileOnDaemon(ctx, pm, profileName, username)
+		if err != nil {
 			return nil, fmt.Errorf("switch profile: %v", err)
 		}
+		return prof, nil
 	}
 
 	activeProf, err := pm.GetActiveProfile()
@@ -251,20 +250,19 @@ func getActiveProfile(ctx context.Context, pm *profilemanager.ProfileManager, pr
 	return activeProf, nil
 }
 
-func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManager, handle string, username string) error {
+func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManager, handle string, username string) (*profilemanager.Profile, error) {
 	resolvedID, err := switchProfile(ctx, handle, username)
 	if err != nil {
-		return fmt.Errorf("switch profile on daemon: %v", err)
+		return nil, fmt.Errorf("switch profile on daemon: %v", err)
 	}
 
 	if err := pm.SwitchProfile(resolvedID); err != nil {
-		return fmt.Errorf("switch profile: %v", err)
+		return nil, fmt.Errorf("switch profile: %v", err)
 	}
 
 	conn, err := DialClientGRPCServer(ctx, daemonAddr)
 	if err != nil {
-		log.Errorf("failed to connect to service CLI interface %v", err)
-		return err
+		return nil, fmt.Errorf("connect to service CLI interface: %w", err)
 	}
 	defer conn.Close()
 
@@ -272,17 +270,17 @@ func switchProfileOnDaemon(ctx context.Context, pm *profilemanager.ProfileManage
 
 	status, err := client.Status(ctx, &proto.StatusRequest{})
 	if err != nil {
-		return fmt.Errorf("unable to get daemon status: %v", err)
+		return nil, fmt.Errorf("unable to get daemon status: %v", err)
 	}
 
 	if status.Status == string(internal.StatusConnected) {
 		if _, err := client.Down(ctx, &proto.DownRequest{}); err != nil {
 			log.Errorf("call service down method: %v", err)
-			return err
+			return nil, err
 		}
 	}
 
-	return nil
+	return &profilemanager.Profile{ID: resolvedID}, nil
 }
 
 // switchProfile asks the daemon to switch to the profile identified by
@@ -345,7 +343,7 @@ func doForegroundLogin(ctx context.Context, cmd *cobra.Command, setupKey string,
 	if err != nil {
 		return fmt.Errorf("foreground login failed: %v", err)
 	}
-	cmd.Println("Logging successfully")
+	cmd.Println("Login successful")
 	return nil
 }
 
@@ -408,7 +406,7 @@ func foregroundGetTokenInfo(ctx context.Context, cmd *cobra.Command, config *pro
 		hint = profileState.Email
 	}
 
-	oAuthFlow, err := auth.NewOAuthFlow(ctx, config, isUnixRunningDesktop(), false, hint)
+	oAuthFlow, err := auth.NewOAuthFlow(ctx, config, util.HasGraphicalSession(), false, hint)
 	if err != nil {
 		return nil, err
 	}
@@ -458,14 +456,6 @@ func openURL(cmd *cobra.Command, verificationURIComplete, userCode string, noBro
 	}
 }
 
-// isUnixRunningDesktop checks if a Linux OS is running desktop environment
-func isUnixRunningDesktop() bool {
-	if runtime.GOOS != "linux" && runtime.GOOS != "freebsd" {
-		return false
-	}
-	return os.Getenv("DESKTOP_SESSION") != "" || os.Getenv("XDG_CURRENT_DESKTOP") != ""
-}
-
 func setEnvAndFlags(cmd *cobra.Command) error {
 	SetFlagsFromEnvVars(rootCmd)
 

client/cmd/logout.go

@@ -3,11 +3,11 @@ package cmd
 import (
 	"context"
 	"fmt"
-	"os/user"
 	"time"
 
 	"github.com/spf13/cobra"
 
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
 	"github.com/netbirdio/netbird/client/proto"
 )
 
@@ -37,7 +37,7 @@ var logoutCmd = &cobra.Command{
 		if profileName != "" {
 			req.ProfileName = &profileName
 
-			currUser, err := user.Current()
+			currUser, err := profilemanager.InvokingUser()
 			if err != nil {
 				return fmt.Errorf("get current user: %v", err)
 			}

client/cmd/profile.go

@@ -4,7 +4,6 @@ import (
 	"context"
 	"errors"
 	"fmt"
-	"os/user"
 	"strings"
 	"text/tabwriter"
 	"time"
@@ -97,7 +96,7 @@ func listProfilesFunc(cmd *cobra.Command, _ []string) error {
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -138,7 +137,7 @@ func addProfileFunc(cmd *cobra.Command, args []string) error {
 		return err
 	}
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -179,7 +178,7 @@ func renameProfileFunc(cmd *cobra.Command, args []string) error {
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -233,7 +232,7 @@ func removeProfileFunc(cmd *cobra.Command, args []string) error {
 	}
 	defer conn.Close()
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}
@@ -261,7 +260,7 @@ func selectProfileFunc(cmd *cobra.Command, args []string) error {
 	profileManager := profilemanager.NewProfileManager()
 	handle := args[0]
 
-	currUser, err := user.Current()
+	currUser, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %w", err)
 	}

client/cmd/root.go

@@ -21,7 +21,9 @@ import (
 	"github.com/spf13/pflag"
 	"google.golang.org/grpc"
 
+	"github.com/netbirdio/netbird/client/anonymize"
 	daddr "github.com/netbirdio/netbird/client/internal/daemonaddr"
+	"github.com/netbirdio/netbird/client/internal/localmetrics"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
 )
 
@@ -30,6 +32,8 @@ const (
 	dnsResolverAddress       = "dns-resolver-address"
 	enableRosenpassFlag      = "enable-rosenpass"
 	rosenpassPermissiveFlag  = "rosenpass-permissive"
+	enableLocalMetricsFlag   = "enable-local-metrics"
+	localMetricsAddressFlag  = "local-metrics-address"
 	preSharedKeyFlag         = "preshared-key"
 	interfaceNameFlag        = "interface-name"
 	wireguardPortFlag        = "wireguard-port"
@@ -69,6 +73,7 @@ var (
 	autoConnectDisabled     bool
 	extraIFaceBlackList     []string
 	anonymizeFlag           bool
+	anonymizeLevelFlag      string
 	dnsRouteInterval        time.Duration
 	// lazyConnEnabled is the parse target for the deprecated --enable-lazy-connection
 	// flag. The flag is inert; the value is no longer read (use NB_LAZY_CONN instead).
@@ -78,6 +83,8 @@ var (
 	updateSettingsDisabled bool
 	captureEnabled         bool
 	networksDisabled       bool
+	localMetricsEnabled    bool
+	localMetricsAddr       string
 
 	rootCmd = &cobra.Command{
 		Use:          "netbird",
@@ -156,7 +163,8 @@ func init() {
 	rootCmd.MarkFlagsMutuallyExclusive("setup-key", "setup-key-file")
 	rootCmd.PersistentFlags().StringVar(&preSharedKey, preSharedKeyFlag, "", "Sets WireGuard PreSharedKey property. If set, then only peers that have the same key can communicate.")
 	rootCmd.PersistentFlags().StringVarP(&hostName, "hostname", "n", "", "Sets a custom hostname for the device")
-	rootCmd.PersistentFlags().BoolVarP(&anonymizeFlag, "anonymize", "A", false, "anonymize IP addresses and non-netbird.io domains in logs and status output")
+	rootCmd.PersistentFlags().BoolVarP(&anonymizeFlag, "anonymize", "A", false, "anonymize public IP addresses, MAC addresses, and non-netbird.io domains in logs and status output; private, CGNAT, and link-local IP ranges are kept (see --anonymize-level strict)")
+	rootCmd.PersistentFlags().StringVar(&anonymizeLevelFlag, "anonymize-level", "", "anonymization level: \"default\" or \"strict\"; strict also anonymizes private, CGNAT, and link-local IP ranges, peer names, and WireGuard public keys. Setting this flag implies --anonymize")
 	rootCmd.PersistentFlags().StringVarP(&configPath, "config", "c", profilemanager.DefaultConfigPath, "Overrides the default profile file location")
 
 	rootCmd.AddCommand(upCmd)
@@ -212,6 +220,8 @@ func init() {
 	upCmd.PersistentFlags().BoolVar(&rosenpassEnabled, enableRosenpassFlag, false, "[Experimental] Enable Rosenpass feature. If enabled, the connection will be post-quantum secured via Rosenpass.")
 	upCmd.PersistentFlags().BoolVar(&rosenpassPermissive, rosenpassPermissiveFlag, false, "[Experimental] Enable Rosenpass in permissive mode to allow this peer to accept WireGuard connections without requiring Rosenpass functionality from peers that do not have Rosenpass enabled.")
 	upCmd.PersistentFlags().BoolVar(&autoConnectDisabled, disableAutoConnectFlag, false, "Disables auto-connect feature. If enabled, then the client won't connect automatically when the service starts.")
+	upCmd.PersistentFlags().BoolVar(&localMetricsEnabled, enableLocalMetricsFlag, false, "Enables a local Prometheus /metrics endpoint exposing connection state (peers, latency, P2P vs relay).")
+	upCmd.PersistentFlags().StringVar(&localMetricsAddr, localMetricsAddressFlag, localmetrics.DefaultListenAddress, "Listen address of the local Prometheus /metrics endpoint.")
 	upCmd.PersistentFlags().BoolVar(&lazyConnEnabled, enableLazyConnectionFlag, false, "Deprecated: no longer used. Lazy connections are controlled by the server and the NB_LAZY_CONN environment variable.")
 	_ = upCmd.PersistentFlags().MarkDeprecated(enableLazyConnectionFlag, "no longer used; lazy connections are controlled by the server and the NB_LAZY_CONN environment variable")
 
@@ -293,6 +303,19 @@ var CLIBackOffSettings = &backoff.ExponentialBackOff{
 	Clock:               backoff.SystemClock,
 }
 
+// effectiveAnonymize resolves the --anonymize and --anonymize-level flags:
+// setting a level implies anonymization, and an invalid level is rejected.
+func effectiveAnonymize() (bool, anonymize.Level, error) {
+	if anonymizeLevelFlag == "" {
+		return anonymizeFlag, anonymize.LevelDefault, nil
+	}
+	level := anonymize.ParseLevel(anonymizeLevelFlag)
+	if !strings.EqualFold(anonymizeLevelFlag, level.String()) {
+		return false, anonymize.LevelDefault, fmt.Errorf("invalid anonymize level %q: use %q or %q", anonymizeLevelFlag, anonymize.LevelDefault.String(), anonymize.LevelStrict.String())
+	}
+	return true, level, nil
+}
+
 func getSetupKey() (string, error) {
 	if setupKeyPath != "" && setupKey == "" {
 		return getSetupKeyFromFile(setupKeyPath)

client/cmd/service_controller.go

@@ -41,13 +41,15 @@ func daemonServerOptions(network string) []grpc.ServerOption {
 	if network == "tcp" {
 		log.Warnf("daemon is listening on TCP (%s): callers carry no verifiable identity over TCP, "+
 			"so privileged operations (SSH root login, SSH auth, enabling the SSH server, management URL changes, "+
-			"deregistration) will be denied. Use a unix socket, or npipe:// on Windows", daemonAddr)
+			"deregistration) will be denied, and the SSH JWT cache is neither filled nor served. "+
+			"Use a unix socket, or npipe:// on Windows", daemonAddr)
 		return nil
 	}
 
-	creds := ipcauth.NewTransportCredentials()
-	if creds == nil {
-		log.Warnf("daemon IPC has no peer-identity primitive on %s: privileged operations will be denied", runtime.GOOS)
+	creds := ipcauth.NewTransportCredentials() //nolint:staticcheck
+	if creds == nil {                          //nolint:staticcheck // nil only on platforms without a peer-identity primitive
+		log.Warnf("daemon IPC has no peer-identity primitive on %s: privileged operations will be denied "+
+			"and the SSH JWT cache is neither filled nor served", runtime.GOOS)
 		return nil
 	}
 

client/cmd/service_socket.go

@@ -27,8 +27,8 @@ func listenOnAddress(addr string) (*socketListener, error) {
 	}
 
 	if network == "npipe" {
-		listener, path, err := listenNamedPipe(address)
-		if err != nil {
+		listener, path, err := listenNamedPipe(address) //nolint:staticcheck
+		if err != nil {                                 //nolint:staticcheck // always errors on non-Windows builds
 			return nil, err
 		}
 		return &socketListener{Listener: listener, network: network, address: path}, nil

client/cmd/status.go

@@ -121,8 +121,14 @@ func statusFunc(cmd *cobra.Command, args []string) error {
 		sessionExpiresAt = ts.AsTime().UTC()
 	}
 
+	anonymizeEnabled, anonymizeLevel, err := effectiveAnonymize()
+	if err != nil {
+		return err
+	}
+
 	var outputInformationHolder = nbstatus.ConvertToStatusOutputOverview(resp.GetFullStatus(), nbstatus.ConvertOptions{
-		Anonymize:            anonymizeFlag,
+		Anonymize:            anonymizeEnabled,
+		AnonymizeLevel:       anonymizeLevel,
 		DaemonVersion:        resp.GetDaemonVersion(),
 		DaemonStatus:         nbstatus.ParseDaemonStatus(status),
 		StatusFilter:         statusFilter,

client/cmd/testutil_test.go

@@ -6,7 +6,7 @@ import (
 	"testing"
 	"time"
 
-	"github.com/golang/mock/gomock"
+	"go.uber.org/mock/gomock"
 	"github.com/stretchr/testify/require"
 	"go.opentelemetry.io/otel"
 	"google.golang.org/grpc"
@@ -124,7 +124,7 @@ func startManagement(t *testing.T, config *config.Config, testFile string) (*grp
 
 	updateManager := update_channel.NewPeersUpdateManager(metrics)
 	requestBuffer := mgmt.NewAccountRequestBuffer(ctx, store)
-	networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.cloud", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersmanager), config)
+	networkMapController := controller.NewController(ctx, store, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.cloud", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(store, peersmanager), config, nil)
 
 	accountManager, err := mgmt.BuildManager(ctx, config, store, networkMapController, jobManager, nil, "", eventStore, nil, false, iv, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManagerMock, false, cacheStore)
 	if err != nil {

client/cmd/up.go

@@ -2,10 +2,10 @@ package cmd
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"net"
 	"net/netip"
-	"os/user"
 	"runtime"
 	"strings"
 	"time"
@@ -21,8 +21,8 @@ import (
 	"github.com/netbirdio/netbird/client/internal"
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
-	"github.com/netbirdio/netbird/client/proto"
 	nbnet "github.com/netbirdio/netbird/client/net"
+	"github.com/netbirdio/netbird/client/proto"
 	"github.com/netbirdio/netbird/client/server"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/shared/management/domain"
@@ -48,6 +48,8 @@ const (
 	profileNameDesc = "profile name to use for the login. If not specified, the last used profile will be used."
 )
 
+var errDaemonActiveProfileUnsupported = errors.New("daemon does not support active profile lookup")
+
 var (
 	foregroundMode     bool
 	dnsLabels          []string
@@ -122,23 +124,25 @@ func upFunc(cmd *cobra.Command, args []string) error {
 
 	pm := profilemanager.NewProfileManager()
 
-	username, err := user.Current()
+	username, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %v", err)
 	}
 
+	var activeProf *profilemanager.Profile
 	var profileSwitched bool
 	// switch profile if provided
 	if profileName != "" {
-		if err := switchOrCreateProfile(cmd.Context(), pm, profileName, username.Username); err != nil {
+		activeProf, err = switchOrCreateProfile(cmd.Context(), pm, profileName, username.Username)
+		if err != nil {
 			return fmt.Errorf("switch profile: %v", err)
 		}
 		profileSwitched = true
-	}
-
-	activeProf, err := pm.GetActiveProfile()
-	if err != nil {
-		return fmt.Errorf("get active profile: %v", err)
+	} else {
+		activeProf, err = pm.GetActiveProfile()
+		if err != nil {
+			return fmt.Errorf("get active profile: %v", err)
+		}
 	}
 
 	if foregroundMode {
@@ -150,13 +154,15 @@ func upFunc(cmd *cobra.Command, args []string) error {
 // switchOrCreateProfile switches the active profile to the one identified by
 // handle, creating it first when it does not exist yet. This restores the
 // pre-0.73 behaviour where `netbird up --profile <name>` auto-creates a
-// missing profile instead of failing.
-func switchOrCreateProfile(ctx context.Context, pm *profilemanager.ProfileManager, handle, username string) error {
+// missing profile instead of failing. Returns the daemon-resolved profile so
+// callers act on it directly instead of re-reading the local state, which is
+// not updated under sudo.
+func switchOrCreateProfile(ctx context.Context, pm *profilemanager.ProfileManager, handle, username string) (*profilemanager.Profile, error) {
 	resolvedID, err := switchProfile(ctx, handle, username)
 	if err != nil {
 		st, ok := gstatus.FromError(err)
 		if !ok || st.Code() != codes.NotFound {
-			return err
+			return nil, err
 		}
 		// Don't fail immediately on a create error: a concurrent run may
 		// have created the profile between the NotFound above and this
@@ -165,16 +171,16 @@ func switchOrCreateProfile(ctx context.Context, pm *profilemanager.ProfileManage
 		_, createErr := createProfile(ctx, handle, username)
 		if resolvedID, err = switchProfile(ctx, handle, username); err != nil {
 			if createErr != nil {
-				return fmt.Errorf("create profile: %w", createErr)
+				return nil, fmt.Errorf("create profile: %w", createErr)
 			}
-			return err
+			return nil, err
 		}
 	}
 
 	if err := pm.SwitchProfile(resolvedID); err != nil {
-		return err
+		return nil, err
 	}
-	return nil
+	return &profilemanager.Profile{ID: resolvedID}, nil
 }
 
 // createProfile dials the daemon and creates a new profile with the given
@@ -302,6 +308,30 @@ func runInDaemonMode(ctx context.Context, cmd *cobra.Command, pm *profilemanager
 		return fmt.Errorf("unable to get daemon status: %v", err)
 	}
 
+	// Under sudo the invoking user's local active-profile mirror is never
+	// written (the SwitchProfile write is a no-op), and plain root has no
+	// invoking user at all — so the mirror read into activeProf above is stale
+	// or defaulted and must not drive the daemon. With no --profile to make the
+	// choice explicit, take the profile the daemon already holds for this user
+	// instead: it stays on the user's current profile rather than silently
+	// switching to the mirror's default, and refuses when the daemon is on
+	// another user's profile.
+	if profileName == "" && !profilemanager.MirrorIsAuthoritative() {
+		u, err := profilemanager.InvokingUser()
+		if err != nil {
+			return fmt.Errorf("get current user: %v", err)
+		}
+		resolved, err := daemonActiveProfileForUser(ctx, client, u.Username)
+		switch {
+		case errors.Is(err, errDaemonActiveProfileUnsupported):
+			log.Warnf("keeping the locally resolved profile: %v", err)
+		case err != nil:
+			return err
+		default:
+			activeProf = resolved
+		}
+	}
+
 	if status.Status == string(internal.StatusConnected) {
 		if !profileSwitched {
 			cmd.Println("Already connected")
@@ -314,7 +344,7 @@ func runInDaemonMode(ctx context.Context, cmd *cobra.Command, pm *profilemanager
 		}
 	}
 
-	username, err := user.Current()
+	username, err := profilemanager.InvokingUser()
 	if err != nil {
 		return fmt.Errorf("get current user: %v", err)
 	}
@@ -398,26 +428,21 @@ func doDaemonUp(ctx context.Context, cmd *cobra.Command, client proto.DaemonServ
 	return nil
 }
 
-func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, profileName, username string) *proto.SetConfigRequest {
-	var req proto.SetConfigRequest
-	req.ProfileName = profileName
-	req.Username = username
-
-	req.ManagementUrl = managementURL
-	req.AdminURL = adminURL
-	req.NatExternalIPs = natExternalIPs
-	req.CustomDNSAddress = customDNSAddressConverted
-	req.ExtraIFaceBlacklist = extraIFaceBlackList
-	req.DnsLabels = dnsLabelsValidated.ToPunycodeList()
-	req.CleanDNSLabels = dnsLabels != nil && len(dnsLabels) == 0
-	req.CleanNATExternalIPs = natExternalIPs != nil && len(natExternalIPs) == 0
-
-	if cmd.Flag(enableRosenpassFlag).Changed {
-		req.RosenpassEnabled = &rosenpassEnabled
-	}
-	if cmd.Flag(rosenpassPermissiveFlag).Changed {
-		req.RosenpassPermissive = &rosenpassPermissive
+// setBoolPtrIfChanged points dst at a copy of val when the named bool flag was
+// explicitly set on cmd. It collapses the repeated
+// "if cmd.Flag(x).Changed { field = &val }" pattern in the request builders into
+// a single call, keeping their cognitive complexity within bounds.
+func setBoolPtrIfChanged(cmd *cobra.Command, name string, dst **bool, val bool) {
+	if cmd.Flag(name).Changed {
+		dst2 := val
+		*dst = &dst2
 	}
+}
+
+// setSSHSetConfigFields copies the SSH server flags the user actually
+// passed into req, leaving the rest unset so the daemon keeps the
+// persisted values.
+func setSSHSetConfigFields(req *proto.SetConfigRequest, cmd *cobra.Command) {
 	if cmd.Flag(serverSSHAllowedFlag).Changed {
 		req.ServerSSHAllowed = &serverSSHAllowed
 	}
@@ -440,6 +465,31 @@ func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, pro
 		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
 		req.SshJWTCacheTTL = &sshJWTCacheTTL32
 	}
+}
+
+func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, profileName, username string) *proto.SetConfigRequest {
+	var req proto.SetConfigRequest
+	req.ProfileName = profileName
+	req.Username = username
+
+	req.ManagementUrl = managementURL
+	req.AdminURL = adminURL
+	req.NatExternalIPs = natExternalIPs
+	req.CustomDNSAddress = customDNSAddressConverted
+	req.ExtraIFaceBlacklist = extraIFaceBlackList
+	req.DnsLabels = dnsLabelsValidated.ToPunycodeList()
+	req.CleanDNSLabels = dnsLabels != nil && len(dnsLabels) == 0
+	req.CleanNATExternalIPs = natExternalIPs != nil && len(natExternalIPs) == 0
+
+	if cmd.Flag(enableRosenpassFlag).Changed {
+		req.RosenpassEnabled = &rosenpassEnabled
+	}
+	if cmd.Flag(rosenpassPermissiveFlag).Changed {
+		req.RosenpassPermissive = &rosenpassPermissive
+	}
+	setSSHSetConfigFields(&req, cmd)
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &req.RemoteJobsAllowed, remoteJobsAllowed)
+
 	if cmd.Flag(interfaceNameFlag).Changed {
 		if err := parseInterfaceName(interfaceName); err != nil {
 			log.Errorf("parse interface name: %v", err)
@@ -499,6 +549,13 @@ func setupSetConfigReq(customDNSAddressConverted []byte, cmd *cobra.Command, pro
 		req.DisableIpv6 = &disableIPv6
 	}
 
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		req.EnableLocalMetrics = &localMetricsEnabled
+	}
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		req.LocalMetricsAddress = &localMetricsAddr
+	}
+
 	return &req
 }
 
@@ -523,6 +580,7 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
 	if cmd.Flag(serverSSHAllowedFlag).Changed {
 		ic.ServerSSHAllowed = &serverSSHAllowed
 	}
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &ic.RemoteJobsAllowed, remoteJobsAllowed)
 
 	if cmd.Flag(enableSSHRootFlag).Changed {
 		ic.EnableSSHRoot = &enableSSHRoot
@@ -616,9 +674,45 @@ func setupConfig(customDNSAddressConverted []byte, cmd *cobra.Command, configFil
 		ic.DisableIPv6 = &disableIPv6
 	}
 
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		ic.LocalMetricsEnabled = &localMetricsEnabled
+	}
+
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		ic.LocalMetricsAddress = &localMetricsAddr
+	}
+
 	return &ic, nil
 }
 
+// setSSHLoginFields copies the SSH server flags the user actually passed
+// into req, leaving the rest unset so the daemon keeps the persisted
+// values.
+func setSSHLoginFields(req *proto.LoginRequest, cmd *cobra.Command) {
+	if cmd.Flag(serverSSHAllowedFlag).Changed {
+		req.ServerSSHAllowed = &serverSSHAllowed
+	}
+	if cmd.Flag(enableSSHRootFlag).Changed {
+		req.EnableSSHRoot = &enableSSHRoot
+	}
+	if cmd.Flag(enableSSHSFTPFlag).Changed {
+		req.EnableSSHSFTP = &enableSSHSFTP
+	}
+	if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
+		req.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
+	}
+	if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
+		req.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
+	}
+	if cmd.Flag(disableSSHAuthFlag).Changed {
+		req.DisableSSHAuth = &disableSSHAuth
+	}
+	if cmd.Flag(sshJWTCacheTTLFlag).Changed {
+		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
+		req.SshJWTCacheTTL = &sshJWTCacheTTL32
+	}
+}
+
 func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte, cmd *cobra.Command) (*proto.LoginRequest, error) {
 	loginRequest := proto.LoginRequest{
 		SetupKey:            providedSetupKey,
@@ -626,7 +720,7 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
 		NatExternalIPs:      natExternalIPs,
 		CleanNATExternalIPs: natExternalIPs != nil && len(natExternalIPs) == 0,
 		CustomDNSAddress:    customDNSAddressConverted,
-		IsUnixDesktopClient: isUnixRunningDesktop(),
+		IsUnixDesktopClient: util.HasGraphicalSession(),
 		Hostname:            hostName,
 		ExtraIFaceBlacklist: extraIFaceBlackList,
 		DnsLabels:           dnsLabels,
@@ -645,39 +739,21 @@ func setupLoginRequest(providedSetupKey string, customDNSAddressConverted []byte
 		loginRequest.RosenpassPermissive = &rosenpassPermissive
 	}
 
-	if cmd.Flag(serverSSHAllowedFlag).Changed {
-		loginRequest.ServerSSHAllowed = &serverSSHAllowed
-	}
-
-	if cmd.Flag(enableSSHRootFlag).Changed {
-		loginRequest.EnableSSHRoot = &enableSSHRoot
-	}
-
-	if cmd.Flag(enableSSHSFTPFlag).Changed {
-		loginRequest.EnableSSHSFTP = &enableSSHSFTP
-	}
-
-	if cmd.Flag(enableSSHLocalPortForwardFlag).Changed {
-		loginRequest.EnableSSHLocalPortForwarding = &enableSSHLocalPortForward
-	}
-
-	if cmd.Flag(enableSSHRemotePortForwardFlag).Changed {
-		loginRequest.EnableSSHRemotePortForwarding = &enableSSHRemotePortForward
-	}
-
-	if cmd.Flag(disableSSHAuthFlag).Changed {
-		loginRequest.DisableSSHAuth = &disableSSHAuth
-	}
-
-	if cmd.Flag(sshJWTCacheTTLFlag).Changed {
-		sshJWTCacheTTL32 := int32(sshJWTCacheTTL)
-		loginRequest.SshJWTCacheTTL = &sshJWTCacheTTL32
-	}
+	setSSHLoginFields(&loginRequest, cmd)
+	setBoolPtrIfChanged(cmd, remoteJobsAllowedFlag, &loginRequest.RemoteJobsAllowed, remoteJobsAllowed)
 
 	if cmd.Flag(disableAutoConnectFlag).Changed {
 		loginRequest.DisableAutoConnect = &autoConnectDisabled
 	}
 
+	if cmd.Flag(enableLocalMetricsFlag).Changed {
+		loginRequest.EnableLocalMetrics = &localMetricsEnabled
+	}
+
+	if cmd.Flag(localMetricsAddressFlag).Changed {
+		loginRequest.LocalMetricsAddress = &localMetricsAddr
+	}
+
 	if cmd.Flag(interfaceNameFlag).Changed {
 		if err := parseInterfaceName(interfaceName); err != nil {
 			return nil, err
@@ -849,3 +925,31 @@ func isValidAddrPort(input string) bool {
 	_, err := netip.ParseAddrPort(input)
 	return err == nil
 }
+
+// daemonActiveProfileForUser returns the profile the daemon currently holds for
+// username, for the no --profile case where the local mirror is not
+// authoritative (sudo or plain root). It returns that profile when the daemon
+// owns it for this user or when the profile is unowned (empty username, as on a
+// fresh install), so the caller acts on the daemon's real state instead of the
+// stale mirror. It denies with a --profile hint when the daemon is on another
+// user's profile, when the lookup fails, or when the daemon reports no active
+// profile. Returns errDaemonActiveProfileUnsupported when the daemon predates
+// the RPC; the caller keeps the mirror-derived profile in that case.
+func daemonActiveProfileForUser(ctx context.Context, client proto.DaemonServiceClient, username string) (*profilemanager.Profile, error) {
+	active, err := client.GetActiveProfile(ctx, &proto.GetActiveProfileRequest{})
+	if err != nil {
+		if st, ok := gstatus.FromError(err); ok && st.Code() == codes.Unimplemented {
+			return nil, fmt.Errorf("%w: %v", errDaemonActiveProfileUnsupported, err)
+		}
+		return nil, fmt.Errorf("pass --profile to choose the profile explicitly: the daemon's active profile could not be verified: %v", err)
+	}
+	if active.GetId() == "" {
+		return nil, fmt.Errorf("pass --profile to choose the profile explicitly: the daemon reported no active profile")
+	}
+	if active.GetUsername() != "" && active.GetUsername() != username {
+		return nil, fmt.Errorf(
+			"pass --profile to choose the profile explicitly: the daemon's active profile is %q (user %q) but this invocation runs for %q",
+			active.GetProfileName(), active.GetUsername(), username)
+	}
+	return &profilemanager.Profile{ID: profilemanager.ID(active.GetId())}, nil
+}

client/cmd/up_test.go

@@ -0,0 +1,88 @@
+package cmd
+
+import (
+	"context"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"google.golang.org/grpc"
+	"google.golang.org/grpc/codes"
+	gstatus "google.golang.org/grpc/status"
+
+	"github.com/netbirdio/netbird/client/internal/profilemanager"
+	"github.com/netbirdio/netbird/client/proto"
+)
+
+type fakeActiveProfileClient struct {
+	proto.DaemonServiceClient
+	resp *proto.GetActiveProfileResponse
+	err  error
+}
+
+func (f *fakeActiveProfileClient) GetActiveProfile(_ context.Context, _ *proto.GetActiveProfileRequest, _ ...grpc.CallOption) (*proto.GetActiveProfileResponse, error) {
+	return f.resp, f.err
+}
+
+func TestDaemonActiveProfileForUserReturnsOwnProfile(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "default", ProfileName: "default", Username: "root"}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.NoError(t, err)
+	require.NotNil(t, prof)
+	assert.Equal(t, profilemanager.ID("default"), prof.ID)
+}
+
+func TestDaemonActiveProfileForUserReturnsUnownedProfile(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "default", ProfileName: "default", Username: ""}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.NoError(t, err)
+	require.NotNil(t, prof)
+	assert.Equal(t, profilemanager.ID("default"), prof.ID)
+}
+
+func TestDaemonActiveProfileForUserKeepsDaemonProfileOverStaleMirror(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "ab12", ProfileName: "work", Username: "misha"}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "misha")
+	require.NoError(t, err)
+	require.NotNil(t, prof)
+	assert.Equal(t, profilemanager.ID("ab12"), prof.ID)
+}
+
+func TestDaemonActiveProfileForUserRejectsOtherUsersProfile(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "ab12", ProfileName: "work", Username: "misha"}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.Error(t, err)
+	assert.Nil(t, prof)
+	assert.Contains(t, err.Error(), "--profile")
+}
+
+func TestDaemonActiveProfileForUserRejectsOtherUsersDefaultProfile(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{Id: "default", ProfileName: "default", Username: "misha"}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.Error(t, err)
+	assert.Nil(t, prof)
+	assert.Contains(t, err.Error(), "--profile")
+}
+
+func TestDaemonActiveProfileForUserRejectsLookupError(t *testing.T) {
+	client := &fakeActiveProfileClient{err: gstatus.Error(codes.Internal, "boom")}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.Error(t, err)
+	assert.Nil(t, prof)
+	assert.Contains(t, err.Error(), "--profile")
+}
+
+func TestDaemonActiveProfileForUserRejectsEmptyResponse(t *testing.T) {
+	client := &fakeActiveProfileClient{resp: &proto.GetActiveProfileResponse{}}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.Error(t, err)
+	assert.Nil(t, prof)
+	assert.Contains(t, err.Error(), "--profile")
+}
+
+func TestDaemonActiveProfileForUserKeepsMirrorWhenDaemonWithoutRPC(t *testing.T) {
+	client := &fakeActiveProfileClient{err: gstatus.Error(codes.Unimplemented, "unknown method")}
+	prof, err := daemonActiveProfileForUser(context.Background(), client, "root")
+	require.ErrorIs(t, err, errDaemonActiveProfileUnsupported)
+	assert.Nil(t, prof)
+}

client/embed/embed.go

@@ -21,7 +21,7 @@ import (
 	"github.com/netbirdio/netbird/client/internal/auth"
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/client/internal/profilemanager"
-	sshcommon "github.com/netbirdio/netbird/client/ssh"
+	nbssh "github.com/netbirdio/netbird/client/ssh"
 	"github.com/netbirdio/netbird/client/system"
 	"github.com/netbirdio/netbird/shared/management/domain"
 	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
@@ -85,12 +85,24 @@ type Options struct {
 	DisableIPv6 bool
 	// BlockInbound blocks all inbound connections from peers
 	BlockInbound bool
+	// EnableRosenpass enables the Rosenpass post-quantum key exchange.
+	EnableRosenpass bool
+	// RosenpassPermissive lets a Rosenpass-enabled peer still connect to peers
+	// that do not run Rosenpass (falling back to the plain WireGuard PSK).
+	RosenpassPermissive bool
 	// BlockLANAccess blocks the embedded peer from reaching the host's
 	// LAN (RFC 1918, link-local, loopback) when it's used as a routing
 	// peer. Mirrors profilemanager.ConfigInput.BlockLANAccess. Useful
 	// when the embedded client must never act as a stepping stone into
 	// the host's local network (e.g. the proxy's overlay peer).
 	BlockLANAccess bool
+	// LazyConnectionEnabled is a tri-state local override for lazy connections,
+	// mirroring the NB_LAZY_CONN env var. Nil defers to the management feature
+	// flag; a set value overrides it in both directions. A short-lived client
+	// that reaches only a few known peers can set this to false, so its peers
+	// connect eagerly and the first request does not wait for the connection to
+	// be established.
+	LazyConnectionEnabled *bool
 	// WireguardPort is the port for the tunnel interface. Use 0 for a random port.
 	WireguardPort *int
 	// MTU is the MTU for the tunnel interface.
@@ -203,6 +215,8 @@ func New(opts Options) (*Client, error) {
 		DisableIPv6:         &opts.DisableIPv6,
 		BlockInbound:        &opts.BlockInbound,
 		BlockLANAccess:      &opts.BlockLANAccess,
+		RosenpassEnabled:    &opts.EnableRosenpass,
+		RosenpassPermissive: &opts.RosenpassPermissive,
 		WireguardPort:       opts.WireguardPort,
 		MTU:                 opts.MTU,
 		DNSLabels:           parsedLabels,
@@ -220,6 +234,15 @@ func New(opts Options) (*Client, error) {
 		config.PrivateKey = opts.PrivateKey
 	}
 
+	if opts.LazyConnectionEnabled != nil {
+		// Runtime-only override, read back through lazyconn.ParseState; a set value
+		// wins over the management feature flag in both directions.
+		config.LazyConnection = "off"
+		if *opts.LazyConnectionEnabled {
+			config.LazyConnection = "on"
+		}
+	}
+
 	if opts.Performance.PreallocatedBuffersPerPool != nil {
 		wgdevice.SetPreallocatedBuffersPerPool(*opts.Performance.PreallocatedBuffersPerPool)
 	}
@@ -521,12 +544,7 @@ func (c *Client) VerifySSHHostKey(peerAddress string, key []byte) error {
 		return err
 	}
 
-	storedKey, found := engine.GetPeerSSHKey(peerAddress)
-	if !found {
-		return sshcommon.ErrPeerNotFound
-	}
-
-	return sshcommon.VerifyHostKey(storedKey, key, peerAddress)
+	return nbssh.PeerKeyLookup(engine.GetPeerSSHKey).VerifySSHHostKey(peerAddress, key)
 }
 
 // SetPerformance retunes a running Client. Only PreallocatedBuffersPerPool

client/embed/embed_test.go

@@ -6,7 +6,7 @@ import (
 	"testing"
 	"time"
 
-	"github.com/golang/mock/gomock"
+	"go.uber.org/mock/gomock"
 	"github.com/stretchr/testify/require"
 	"google.golang.org/grpc"
 
@@ -146,7 +146,7 @@ func startManagement(t *testing.T, signalAddr string) string {
 
 	updateManager := update_channel.NewPeersUpdateManager(metrics)
 	requestBuffer := mgmt.NewAccountRequestBuffer(context.Background(), testStore)
-	networkMapController := controller.NewController(context.Background(), testStore, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(testStore, peersManager), cfg)
+	networkMapController := controller.NewController(context.Background(), testStore, metrics, updateManager, requestBuffer, mgmt.MockIntegratedValidator{}, settingsMockManager, "netbird.selfhosted", port_forwarding.NewControllerMock(), manager.NewEphemeralManager(testStore, peersManager), cfg, nil)
 	accountManager, err := mgmt.BuildManager(context.Background(), cfg, testStore, networkMapController, jobManager, nil, "", eventStore, nil, false, iv, metrics, port_forwarding.NewControllerMock(), settingsMockManager, permissionsManager, false, cacheStore)
 	require.NoError(t, err)
 

client/firewall/firewalld/firewalld.go

@@ -2,8 +2,8 @@
 // its wg interface into firewalld's "trusted" zone. This is required because
 // firewalld's nftables chains are created with NFT_CHAIN_OWNER on recent
 // versions, which returns EPERM to any other process that tries to insert
-// rules into them. The workaround mirrors what Tailscale does: let firewalld
-// itself add the accept rules to its own chains by trusting the interface.
+// rules into them. Trusting the interface makes firewalld itself add the
+// accept rules to its own chains instead.
 package firewalld
 
 // TrustedZone is the firewalld zone name used for interfaces whose traffic

client/firewall/iptables/acl_linux.go

@@ -42,6 +42,7 @@ type aclManager struct {
 	optionalEntries map[string][]entry
 	ipsetStore      *ipsetStore
 	v6              bool
+	ipsetSupported  bool
 
 	stateManager *statemanager.Manager
 }
@@ -60,6 +61,8 @@ func newAclManager(iptablesClient *iptables.IPTables, wgIface iFaceMapper) (*acl
 func (m *aclManager) init(stateManager *statemanager.Manager) error {
 	m.stateManager = stateManager
 
+	m.ipsetSupported = m.probeIPSetSupport()
+
 	m.seedInitialEntries()
 	m.seedInitialOptionalEntries()
 
@@ -91,6 +94,12 @@ func (m *aclManager) AddPeerFiltering(
 	if m.v6 && ipsetName != "" {
 		ipsetName += "-v6"
 	}
+	// When the kernel lacks the required ipset hash module, fall back to
+	// per-IP iptables rules (pre-0.68 behavior) so ACLs keep working instead
+	// of silently leaving the chain empty.
+	if ipsetName != "" && !m.ipsetSupported {
+		ipsetName = ""
+	}
 	proto := protoForFamily(protocol, m.v6)
 	specs := filterRuleSpecs(ip, proto, sPort, dPort, action, ipsetName)
 
@@ -498,6 +507,40 @@ func transformIPsetName(ipsetName string, sPort, dPort *firewall.Port, action fi
 	}
 }
 
+// probeIPSetSupport checks whether the kernel can create the ipset type used for
+// ACL rules. On kernels lacking the required ipset hash module, ipset creation
+// fails (e.g. "invalid argument"), which would otherwise leave the ACL chain
+// empty and silently drop all policy-permitted inbound traffic. When unsupported,
+// the manager falls back to per-IP iptables rules.
+func (m *aclManager) probeIPSetSupport() bool {
+	// Use a unique name so concurrent processes don't collide and we only ever
+	// destroy the set we created ourselves. ipset names are limited to 31 chars,
+	// so use a short random suffix.
+	probeName := "nb-probe-" + uuid.New().String()[:8]
+
+	opts := ipset.CreateOptions{
+		Replace: true,
+	}
+	if m.v6 {
+		opts.Family = ipset.FamilyIPV6
+	}
+
+	if err := ipset.Create(probeName, ipset.TypeHashNet, opts); err != nil {
+		log.Warnf("ipset is not available (failed to create probe set: %v); "+
+			"falling back to per-IP iptables ACL rules. Ensure the kernel provides "+
+			"the ipset hash:net module (ip_set_hash_net) for better performance with large rule sets", err)
+		return false
+	}
+
+	defer func() {
+		if err := ipset.Destroy(probeName); err != nil {
+			log.Debugf("destroy ipset probe set %q: %v", probeName, err)
+		}
+	}()
+
+	return true
+}
+
 func (m *aclManager) createIPSet(name string) error {
 	opts := ipset.CreateOptions{
 		Replace: true,

client/firewall/iptables/dnat_refcount_linux_test.go

@@ -0,0 +1,240 @@
+//go:build privileged
+
+package iptables
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	fw "github.com/netbirdio/netbird/client/firewall/manager"
+	"github.com/netbirdio/netbird/client/iface"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+)
+
+func iptRefcountIfaceV4() *iFaceMock {
+	return &iFaceMock{
+		NameFunc: func() string { return "wt-refcount" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("10.20.0.1"),
+				Network: netip.MustParsePrefix("10.20.0.0/24"),
+			}
+		},
+	}
+}
+
+func iptRefcountIfaceDual() *iFaceMock {
+	return &iFaceMock{
+		NameFunc: func() string { return "wt-refcount" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("10.20.0.1"),
+				Network: netip.MustParsePrefix("10.20.0.0/24"),
+				IPv6:    netip.MustParseAddr("fd00::1"),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			}
+		},
+	}
+}
+
+func newIptRefcountManager(t *testing.T, dual bool) *Manager {
+	t.Helper()
+	var ifMock *iFaceMock
+	if dual {
+		ifMock = iptRefcountIfaceDual()
+	} else {
+		ifMock = iptRefcountIfaceV4()
+	}
+	m, err := Create(ifMock, iface.DefaultMTU)
+	require.NoError(t, err, "create manager")
+	require.NoError(t, m.Init(nil), "init manager")
+	t.Cleanup(func() {
+		require.NoError(t, m.Close(nil), "close manager")
+	})
+	return m
+}
+
+func iptDnatV4(port uint16) fw.ForwardRule {
+	return fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{port}},
+		TranslatedAddress: netip.MustParseAddr("10.20.0.2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+}
+
+func iptDnatV6(port uint16) fw.ForwardRule {
+	return fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{port}},
+		TranslatedAddress: netip.MustParseAddr("fd00::2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+}
+
+// TestIptablesRouting_RepeatedEnableSingleReference verifies that EnableRouting
+// (called on every network-map update) holds at most one reference per family
+// and a single DisableRouting drops both back to zero.
+func TestIptablesRouting_RepeatedEnableSingleReference(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	require.NoError(t, m.EnableRouting(), "first enable")
+	require.NoError(t, m.EnableRouting(), "second enable")
+	require.NoError(t, m.EnableRouting(), "third enable")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 1, v4, "repeated enable holds a single v4 reference")
+	assert.Equal(t, 1, v6, "repeated enable holds a single v6 reference")
+
+	require.NoError(t, m.DisableRouting(), "disable")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "single disable releases the v4 reference")
+	assert.Equal(t, 0, v6, "single disable releases the v6 reference")
+}
+
+// TestIptablesRouting_DisableKeepsDNATReference verifies that an unpaired
+// DisableRouting does not release references held by active DNAT rules.
+func TestIptablesRouting_DisableKeepsDNATReference(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	r1, err := m.AddDNATRule(iptDnatV6(9095))
+	require.NoError(t, err, "add v6 dnat")
+
+	require.NoError(t, m.DisableRouting(), "unpaired disable")
+	_, v6 := state.Counts()
+	assert.Equal(t, 1, v6, "DNAT-held reference survives unpaired DisableRouting")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6, "delete releases the DNAT reference")
+}
+
+// TestIptablesDNAT_RefcountBalancedV4 covers a Balanced Add/Delete pair on v4.
+func TestIptablesDNAT_RefcountBalancedV4(t *testing.T) {
+	m := newIptRefcountManager(t, false)
+	state := m.router.ipFwdState
+
+	r1, err := m.AddDNATRule(iptDnatV4(7081))
+	require.NoError(t, err, "add v4 dnat 1")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 1, v4, "v4 refcount after first add")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	r2, err := m.AddDNATRule(iptDnatV4(7082))
+	require.NoError(t, err, "add v4 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 2, v4, "v4 refcount after second add")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	require.NoError(t, m.DeleteDNATRule(r1))
+	v4, v6 = state.Counts()
+	assert.Equal(t, 1, v4, "v4 refcount after first delete")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	require.NoError(t, m.DeleteDNATRule(r2))
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount after second delete")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+}
+
+// TestIptablesDNAT_RefcountBalancedV6 checks the v6 path increments v6 only and
+// decrements back to zero.
+func TestIptablesDNAT_RefcountBalancedV6(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	require.NotNil(t, m.router6, "v6 router")
+	require.Same(t, m.router.ipFwdState, m.router6.ipFwdState, "shared state")
+	state := m.router.ipFwdState
+
+	r1, err := m.AddDNATRule(iptDnatV6(9081))
+	require.NoError(t, err, "add v6 dnat 1")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 1, v6, "v6 refcount after first add")
+
+	r2, err := m.AddDNATRule(iptDnatV6(9082))
+	require.NoError(t, err, "add v6 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unchanged")
+	assert.Equal(t, 2, v6, "v6 refcount after second add")
+
+	require.NoError(t, m.DeleteDNATRule(r1))
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unchanged")
+	assert.Equal(t, 1, v6, "v6 refcount after first delete")
+
+	require.NoError(t, m.DeleteDNATRule(r2))
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6, "v6 refcount after second delete")
+}
+
+// TestIptablesDNAT_DuplicateAddNoLeak verifies the duplicate-rule path returns
+// without bumping the refcount.
+func TestIptablesDNAT_DuplicateAddNoLeak(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	rule := iptDnatV4(7083)
+	r1, err := m.AddDNATRule(rule)
+	require.NoError(t, err)
+	v4, _ := state.Counts()
+	assert.Equal(t, 1, v4)
+
+	_, err = m.AddDNATRule(rule)
+	require.NoError(t, err, "duplicate add")
+	v4, _ = state.Counts()
+	assert.Equal(t, 1, v4, "duplicate add must not increment")
+
+	require.NoError(t, m.DeleteDNATRule(r1))
+	v4, _ = state.Counts()
+	assert.Equal(t, 0, v4, "single delete must drop to zero")
+}
+
+// TestIptablesDNAT_DeleteMissingNoUnderflow verifies Delete on an unknown rule
+// neither errors nor releases the refcount.
+func TestIptablesDNAT_DeleteMissingNoUnderflow(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	phantom := iptDnatV4(7099)
+	require.NoError(t, m.DeleteDNATRule(&phantom), "delete missing v4")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6)
+
+	phantom6 := iptDnatV6(9099)
+	require.NoError(t, m.DeleteDNATRule(&phantom6), "delete missing v6")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6)
+
+	r1, err := m.AddDNATRule(iptDnatV4(7100))
+	require.NoError(t, err)
+	v4, _ = state.Counts()
+	assert.Equal(t, 1, v4, "real add still increments after phantom delete")
+	require.NoError(t, m.DeleteDNATRule(r1))
+}
+
+// TestIptablesDNAT_DoubleDeleteNoUnderflow verifies a second Delete on the same
+// rule is a no-op.
+func TestIptablesDNAT_DoubleDeleteNoUnderflow(t *testing.T) {
+	m := newIptRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	r1, err := m.AddDNATRule(iptDnatV6(9083))
+	require.NoError(t, err)
+	_, v6 := state.Counts()
+	assert.Equal(t, 1, v6)
+
+	require.NoError(t, m.DeleteDNATRule(r1), "first delete")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6)
+
+	require.NoError(t, m.DeleteDNATRule(r1), "second delete must be no-op")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6, "double delete must not underflow")
+}

client/firewall/iptables/manager_linux.go

@@ -89,7 +89,7 @@ func (m *Manager) createIPv6Components(wgIface iFaceMapper, mtu uint16) error {
 	}
 
 	// Share the same IP forwarding state with the v4 router, since
-	// EnableIPForwarding controls both v4 and v6 sysctls.
+	// Forwarding refcounter is per-family but shared between v4 and v6 routers.
 	m.router6.ipFwdState = m.router.ipFwdState
 
 	m.aclMgr6, err = newAclManager(ip6Client, wgIface)
@@ -402,17 +402,12 @@ func (m *Manager) SetLogLevel(log.Level) {
 }
 
 func (m *Manager) EnableRouting() error {
-	if err := m.router.ipFwdState.RequestForwarding(); err != nil {
-		return fmt.Errorf("enable IP forwarding: %w", err)
-	}
-	return nil
+	// v6 only when the overlay actually has v6.
+	return m.router.ipFwdState.RequestRouting(m.router6 != nil)
 }
 
 func (m *Manager) DisableRouting() error {
-	if err := m.router.ipFwdState.ReleaseForwarding(); err != nil {
-		return fmt.Errorf("disable IP forwarding: %w", err)
-	}
-	return nil
+	return m.router.ipFwdState.ReleaseRouting()
 }
 
 // AddDNATRule adds a DNAT rule

client/firewall/iptables/manager_linux_test.go

@@ -291,3 +291,40 @@ func TestIptablesCreatePerformance(t *testing.T) {
 		})
 	}
 }
+
+// TestIptablesACLIPSetFallback verifies that when the kernel lacks ipset support,
+// the ACL manager falls back to per-IP iptables rules (-s <ip>) instead of
+// silently leaving the chain empty. See discussion #6125.
+func TestIptablesACLIPSetFallback(t *testing.T) {
+	ipv4Client, err := iptables.NewWithProtocol(iptables.ProtocolIPv4)
+	require.NoError(t, err)
+
+	// Use Create()/Init() so the router-owned chains (chainRTFWDIN/OUT) are
+	// created before the ACL manager's createDefaultChains() references them.
+	manager, err := Create(ifaceMock, iface.DefaultMTU)
+	require.NoError(t, err)
+	require.NoError(t, manager.Init(nil))
+
+	aclMgr := manager.aclMgr
+	// Simulate a kernel without the ipset hash module.
+	aclMgr.ipsetSupported = false
+
+	defer func() {
+		require.NoError(t, manager.Close(nil))
+	}()
+
+	ip := netip.MustParseAddr("10.20.0.42")
+	port := &fw.Port{Values: []uint16{22}}
+
+	rules, err := aclMgr.AddPeerFiltering(nil, ip.AsSlice(), "tcp", nil, port, fw.ActionAccept, "nb0000001")
+	require.NoError(t, err, "AddPeerFiltering should succeed via fallback")
+	require.NotEmpty(t, rules)
+
+	rule := rules[0].(*Rule)
+	require.Empty(t, rule.ipsetName, "fallback rule must not reference an ipset")
+	require.Contains(t, strings.Join(rule.specs, " "), "-s 10.20.0.42", "fallback rule must match by source IP")
+	require.NotContains(t, strings.Join(rule.specs, " "), "--match-set", "fallback rule must not use ipset matching")
+
+	// The rule must actually be present in the ACL chain (not silently dropped).
+	checkRuleSpecs(t, ipv4Client, rule.chain, true, rule.specs...)
+}

client/firewall/iptables/router_linux.go

@@ -102,7 +102,7 @@ func newRouter(iptablesClient *iptables.IPTables, wgIface iFaceMapper, mtu uint1
 		wgIface:        wgIface,
 		mtu:            mtu,
 		v6:             iptablesClient.Proto() == iptables.ProtocolIPv6,
-		ipFwdState:     ipfwdstate.NewIPForwardingState(),
+		ipFwdState:     ipfwdstate.NewIPForwardingState(wgIface.Name()),
 	}
 
 	r.ipsetCounter = refcounter.New(
@@ -770,10 +770,6 @@ func (r *router) updateState() {
 }
 
 func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
-	if err := r.ipFwdState.RequestForwarding(); err != nil {
-		return nil, err
-	}
-
 	ruleKey := rule.ID()
 	if _, exists := r.rules[ruleKey+dnatSuffix]; exists {
 		return rule, nil
@@ -840,18 +836,34 @@ func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
 
 	for key, ruleInfo := range rules {
 		if err := r.iptablesClient.Append(ruleInfo.table, ruleInfo.chain, ruleInfo.rule...); err != nil {
-			if rollbackErr := r.rollbackRules(rules); rollbackErr != nil {
-				log.Errorf("rollback failed: %v", rollbackErr)
-			}
+			r.cleanupFailedDNATAdd(rules)
 			return nil, fmt.Errorf("add rule %s: %w", key, err)
 		}
 		r.rules[key] = ruleInfo.rule
 	}
 
+	if err := r.ipFwdState.RequestForwarding(r.v6); err != nil {
+		r.cleanupFailedDNATAdd(rules)
+		return nil, fmt.Errorf("enable forwarding: %w", err)
+	}
+
 	r.updateState()
 	return rule, nil
 }
 
+// cleanupFailedDNATAdd removes the bookkeeping written by a partially applied
+// AddDNATRule before rolling back the kernel rules, so no entries remain that
+// never got a forwarding refcount. rollbackRules re-adds entries it failed to
+// remove from the kernel.
+func (r *router) cleanupFailedDNATAdd(rules map[string]ruleInfo) {
+	for key := range rules {
+		delete(r.rules, key)
+	}
+	if err := r.rollbackRules(rules); err != nil {
+		log.Errorf("rollback failed: %v", err)
+	}
+}
+
 func (r *router) rollbackRules(rules map[string]ruleInfo) error {
 	var merr *multierror.Error
 	for key, ruleInfo := range rules {
@@ -868,32 +880,47 @@ func (r *router) rollbackRules(rules map[string]ruleInfo) error {
 }
 
 func (r *router) DeleteDNATRule(rule firewall.Rule) error {
-	if err := r.ipFwdState.ReleaseForwarding(); err != nil {
-		log.Errorf("%v", err)
-	}
-
 	ruleKey := rule.ID()
 
+	_, hadDNAT := r.rules[ruleKey+dnatSuffix]
+	_, hadSNAT := r.rules[ruleKey+snatSuffix]
+	_, hadFWD := r.rules[ruleKey+fwdSuffix]
+	if !hadDNAT && !hadSNAT && !hadFWD {
+		return nil
+	}
+
 	var merr *multierror.Error
 	if dnatRule, exists := r.rules[ruleKey+dnatSuffix]; exists {
 		if err := r.iptablesClient.Delete(tableNat, chainRTRDR, dnatRule...); err != nil {
 			merr = multierror.Append(merr, fmt.Errorf("delete DNAT rule: %w", err))
+		} else {
+			delete(r.rules, ruleKey+dnatSuffix)
 		}
-		delete(r.rules, ruleKey+dnatSuffix)
 	}
 
 	if snatRule, exists := r.rules[ruleKey+snatSuffix]; exists {
 		if err := r.iptablesClient.Delete(tableNat, chainRTNAT, snatRule...); err != nil {
 			merr = multierror.Append(merr, fmt.Errorf("delete SNAT rule: %w", err))
+		} else {
+			delete(r.rules, ruleKey+snatSuffix)
 		}
-		delete(r.rules, ruleKey+snatSuffix)
 	}
 
 	if fwdRule, exists := r.rules[ruleKey+fwdSuffix]; exists {
 		if err := r.iptablesClient.Delete(tableFilter, chainRTFWDOUT, fwdRule...); err != nil {
 			merr = multierror.Append(merr, fmt.Errorf("delete forward rule: %w", err))
+		} else {
+			delete(r.rules, ruleKey+fwdSuffix)
+		}
+	}
+
+	// Release the refcount only once all rules are gone from the kernel. On
+	// partial failure the failed entries stay in r.rules so a retry can remove
+	// them and release then.
+	if merr == nil {
+		if err := r.ipFwdState.ReleaseForwarding(r.v6); err != nil {
+			log.Errorf("%v", err)
 		}
-		delete(r.rules, ruleKey+fwdSuffix)
 	}
 
 	r.updateState()

client/firewall/nftables/dnat_refcount_linux_test.go

@@ -0,0 +1,249 @@
+//go:build privileged
+
+package nftables
+
+import (
+	"net/netip"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	fw "github.com/netbirdio/netbird/client/firewall/manager"
+	"github.com/netbirdio/netbird/client/iface"
+	"github.com/netbirdio/netbird/client/iface/wgaddr"
+)
+
+func nftRefcountIfaceV4() *iFaceMock {
+	return &iFaceMock{
+		NameFunc: func() string { return "wt-refcount" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("100.96.0.1"),
+				Network: netip.MustParsePrefix("100.96.0.0/16"),
+			}
+		},
+	}
+}
+
+func nftRefcountIfaceDual() *iFaceMock {
+	return &iFaceMock{
+		NameFunc: func() string { return "wt-refcount" },
+		AddressFunc: func() wgaddr.Address {
+			return wgaddr.Address{
+				IP:      netip.MustParseAddr("100.96.0.1"),
+				Network: netip.MustParsePrefix("100.96.0.0/16"),
+				IPv6:    netip.MustParseAddr("fd00::1"),
+				IPv6Net: netip.MustParsePrefix("fd00::/64"),
+			}
+		},
+	}
+}
+
+func newNftRefcountManager(t *testing.T, dual bool) *Manager {
+	t.Helper()
+	if check() != NFTABLES {
+		t.Skip("nftables not supported on this system")
+	}
+	var ifMock *iFaceMock
+	if dual {
+		ifMock = nftRefcountIfaceDual()
+	} else {
+		ifMock = nftRefcountIfaceV4()
+	}
+	m, err := Create(ifMock, iface.DefaultMTU)
+	require.NoError(t, err, "create manager")
+	require.NoError(t, m.Init(nil), "init manager")
+	t.Cleanup(func() {
+		require.NoError(t, m.Close(nil), "close manager")
+	})
+	return m
+}
+
+func dnatV4(port uint16) fw.ForwardRule {
+	return fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{port}},
+		TranslatedAddress: netip.MustParseAddr("100.96.0.2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+}
+
+func dnatV6(port uint16) fw.ForwardRule {
+	return fw.ForwardRule{
+		Protocol:          fw.ProtocolTCP,
+		DestinationPort:   fw.Port{Values: []uint16{port}},
+		TranslatedAddress: netip.MustParseAddr("fd00::2"),
+		TranslatedPort:    fw.Port{Values: []uint16{80}},
+	}
+}
+
+// TestNftablesDNAT_RefcountBalancedV4 verifies that Add/Delete pairs leave the
+// v4 refcount at zero.
+func TestNftablesDNAT_RefcountBalancedV4(t *testing.T) {
+	m := newNftRefcountManager(t, false)
+	state := m.router.ipFwdState
+
+	r1, err := m.AddDNATRule(dnatV4(8081))
+	require.NoError(t, err, "add v4 dnat 1")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 1, v4, "v4 refcount after first add")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	r2, err := m.AddDNATRule(dnatV4(8082))
+	require.NoError(t, err, "add v4 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 2, v4, "v4 refcount after second add")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v4 dnat 1")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 1, v4, "v4 refcount after first delete")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+
+	require.NoError(t, m.DeleteDNATRule(r2), "delete v4 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount after second delete")
+	assert.Equal(t, 0, v6, "v6 refcount unchanged")
+}
+
+// TestNftablesDNAT_RefcountBalancedV6 verifies the v6 path increments v6 only
+// and decrements back to zero on Delete.
+func TestNftablesDNAT_RefcountBalancedV6(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	require.NotNil(t, m.router6, "v6 router")
+	require.Same(t, m.router.ipFwdState, m.router6.ipFwdState, "shared state")
+	state := m.router.ipFwdState
+
+	r1, err := m.AddDNATRule(dnatV6(9091))
+	require.NoError(t, err, "add v6 dnat 1")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unchanged")
+	assert.Equal(t, 1, v6, "v6 refcount after first add")
+
+	r2, err := m.AddDNATRule(dnatV6(9092))
+	require.NoError(t, err, "add v6 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 2, v6, "v6 refcount after second add")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat 1")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unchanged")
+	assert.Equal(t, 1, v6, "v6 refcount after first delete")
+
+	require.NoError(t, m.DeleteDNATRule(r2), "delete v6 dnat 2")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6, "v6 refcount after second delete")
+}
+
+// TestNftablesDNAT_DuplicateAddNoLeak verifies that a duplicate Add (same
+// ForwardRule) does not double-increment the refcount.
+func TestNftablesDNAT_DuplicateAddNoLeak(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	rule := dnatV4(8083)
+	r1, err := m.AddDNATRule(rule)
+	require.NoError(t, err, "add v4 dnat")
+	v4, _ := state.Counts()
+	assert.Equal(t, 1, v4)
+
+	// duplicate add: same rule ID, must be a no-op for the refcount.
+	_, err = m.AddDNATRule(rule)
+	require.NoError(t, err, "duplicate add")
+	v4, _ = state.Counts()
+	assert.Equal(t, 1, v4, "duplicate add must not increment")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v4 dnat")
+	v4, _ = state.Counts()
+	assert.Equal(t, 0, v4, "single delete must drop to zero")
+}
+
+// TestNftablesDNAT_DeleteMissingNoUnderflow verifies deleting a rule that was
+// never added does not underflow the refcount.
+func TestNftablesDNAT_DeleteMissingNoUnderflow(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	// Construct a Rule reference for something never added. The router stores
+	// rules by ID(), and DeleteDNATRule looks them up in r.rules; a missing
+	// entry must be a no-op rather than calling Release.
+	phantom := dnatV4(8099)
+	require.NoError(t, m.DeleteDNATRule(&phantom), "delete missing v4 dnat")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 0, v4, "v4 refcount unaffected by missing delete")
+	assert.Equal(t, 0, v6, "v6 refcount unaffected")
+
+	phantom6 := dnatV6(9099)
+	require.NoError(t, m.DeleteDNATRule(&phantom6), "delete missing v6 dnat")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4)
+	assert.Equal(t, 0, v6, "v6 refcount unaffected by missing delete")
+
+	// And after a phantom delete, a real add still results in count=1.
+	r1, err := m.AddDNATRule(dnatV4(8100))
+	require.NoError(t, err, "add v4 dnat after phantom delete")
+	v4, _ = state.Counts()
+	assert.Equal(t, 1, v4, "real add still increments after phantom delete")
+	require.NoError(t, m.DeleteDNATRule(r1))
+}
+
+// TestNftablesRouting_RepeatedEnableSingleReference verifies that EnableRouting
+// (called on every network-map update) holds at most one reference per family
+// and a single DisableRouting drops both back to zero.
+func TestNftablesRouting_RepeatedEnableSingleReference(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	require.NoError(t, m.EnableRouting(), "first enable")
+	require.NoError(t, m.EnableRouting(), "second enable")
+	require.NoError(t, m.EnableRouting(), "third enable")
+	v4, v6 := state.Counts()
+	assert.Equal(t, 1, v4, "repeated enable holds a single v4 reference")
+	assert.Equal(t, 1, v6, "repeated enable holds a single v6 reference")
+
+	require.NoError(t, m.DisableRouting(), "disable")
+	v4, v6 = state.Counts()
+	assert.Equal(t, 0, v4, "single disable releases the v4 reference")
+	assert.Equal(t, 0, v6, "single disable releases the v6 reference")
+}
+
+// TestNftablesRouting_DisableKeepsDNATReference verifies that an unpaired
+// DisableRouting does not release references held by active DNAT rules.
+func TestNftablesRouting_DisableKeepsDNATReference(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	r1, err := m.AddDNATRule(dnatV6(9095))
+	require.NoError(t, err, "add v6 dnat")
+
+	require.NoError(t, m.DisableRouting(), "unpaired disable")
+	_, v6 := state.Counts()
+	assert.Equal(t, 1, v6, "DNAT-held reference survives unpaired DisableRouting")
+
+	require.NoError(t, m.DeleteDNATRule(r1), "delete v6 dnat")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6, "delete releases the DNAT reference")
+}
+
+// TestNftablesDNAT_DoubleDeleteNoUnderflow verifies that deleting the same rule
+// twice does not underflow the refcount (the second delete is a no-op).
+func TestNftablesDNAT_DoubleDeleteNoUnderflow(t *testing.T) {
+	m := newNftRefcountManager(t, true)
+	state := m.router.ipFwdState
+
+	r1, err := m.AddDNATRule(dnatV6(9093))
+	require.NoError(t, err)
+	_, v6 := state.Counts()
+	assert.Equal(t, 1, v6)
+
+	require.NoError(t, m.DeleteDNATRule(r1), "first delete")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6)
+
+	require.NoError(t, m.DeleteDNATRule(r1), "second delete must be no-op")
+	_, v6 = state.Counts()
+	assert.Equal(t, 0, v6, "double delete must not underflow")
+}

client/firewall/nftables/manager_linux.go

@@ -105,8 +105,8 @@ func (m *Manager) createIPv6Components(tableName string, wgIface iFaceMapper, mt
 		return fmt.Errorf("create v6 router: %w", err)
 	}
 
-	// Share the same IP forwarding state with the v4 router, since
-	// EnableIPForwarding controls both v4 and v6 sysctls.
+	// Share the per-family forwarding refcounter with the v4 router so a v4
+	// rule and a v6 rule against the same state machine cooperate cleanly.
 	m.router6.ipFwdState = m.router.ipFwdState
 
 	m.aclManager6, err = newAclManager(workTable6, wgIface, chainNameRoutingFw)
@@ -530,17 +530,12 @@ func (m *Manager) SetLogLevel(log.Level) {
 }
 
 func (m *Manager) EnableRouting() error {
-	if err := m.router.ipFwdState.RequestForwarding(); err != nil {
-		return fmt.Errorf("enable IP forwarding: %w", err)
-	}
-	return nil
+	// v6 only when the overlay actually has v6.
+	return m.router.ipFwdState.RequestRouting(m.router6 != nil)
 }
 
 func (m *Manager) DisableRouting() error {
-	if err := m.router.ipFwdState.ReleaseForwarding(); err != nil {
-		return fmt.Errorf("disable IP forwarding: %w", err)
-	}
-	return nil
+	return m.router.ipFwdState.ReleaseRouting()
 }
 
 // Flush rule/chain/set operations from the buffer

client/firewall/nftables/router_linux.go

@@ -93,7 +93,7 @@ func newRouter(workTable *nftables.Table, wgIface iFaceMapper, mtu uint16) (*rou
 		rules:      make(map[string]*nftables.Rule),
 		af:         familyForAddr(workTable.Family == nftables.TableFamilyIPv4),
 		wgIface:    wgIface,
-		ipFwdState: ipfwdstate.NewIPForwardingState(),
+		ipFwdState: ipfwdstate.NewIPForwardingState(wgIface.Name()),
 		mtu:        mtu,
 	}
 
@@ -763,7 +763,7 @@ func (r *router) addNatRule(pair firewall.RouterPair) error {
 	exprs = append(exprs, sourceExp...)
 	exprs = append(exprs, destExp...)
 
-	var markValue uint32 = nbnet.PreroutingFwmarkMasquerade
+	markValue := nbnet.PreroutingFwmarkMasquerade
 	if pair.Inverse {
 		markValue = nbnet.PreroutingFwmarkMasqueradeReturn
 	}
@@ -1553,10 +1553,6 @@ func (r *router) refreshRulesMap() error {
 }
 
 func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
-	if err := r.ipFwdState.RequestForwarding(); err != nil {
-		return nil, err
-	}
-
 	ruleKey := rule.ID()
 	if _, exists := r.rules[ruleKey+dnatSuffix]; exists {
 		return rule, nil
@@ -1567,7 +1563,18 @@ func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
 		return nil, fmt.Errorf("convert protocol to number: %w", err)
 	}
 
+	// Request forwarding before queueing rules: addDnatRedirect/addDnatMasq
+	// buffer netlink messages on r.conn that the next caller's Flush would
+	// commit if we returned without flushing them ourselves.
+	v6 := r.af.tableFamily == nftables.TableFamilyIPv6
+	if err := r.ipFwdState.RequestForwarding(v6); err != nil {
+		return nil, fmt.Errorf("enable forwarding: %w", err)
+	}
+
 	if err := r.addDnatRedirect(rule, protoNum, ruleKey); err != nil {
+		if rerr := r.ipFwdState.ReleaseForwarding(v6); rerr != nil {
+			log.Warnf("rollback forwarding refcount: %v", rerr)
+		}
 		return nil, err
 	}
 
@@ -1579,6 +1586,11 @@ func (r *router) AddDNATRule(rule firewall.ForwardRule) (firewall.Rule, error) {
 	// TODO: find chains with drop policies and add rules there
 
 	if err := r.conn.Flush(); err != nil {
+		if rerr := r.ipFwdState.ReleaseForwarding(v6); rerr != nil {
+			log.Warnf("rollback forwarding refcount: %v", rerr)
+		}
+		delete(r.rules, ruleKey+dnatSuffix)
+		delete(r.rules, ruleKey+snatSuffix)
 		return nil, fmt.Errorf("flush rules: %w", err)
 	}
 
@@ -1781,16 +1793,18 @@ func (r *router) addDnatMasq(rule firewall.ForwardRule, protoNum uint8, ruleKey
 }
 
 func (r *router) DeleteDNATRule(rule firewall.Rule) error {
-	if err := r.ipFwdState.ReleaseForwarding(); err != nil {
-		log.Errorf("%v", err)
-	}
-
 	ruleKey := rule.ID()
 
 	if err := r.refreshRulesMap(); err != nil {
 		return fmt.Errorf(refreshRulesMapError, err)
 	}
 
+	_, hadDNAT := r.rules[ruleKey+dnatSuffix]
+	_, hadSNAT := r.rules[ruleKey+snatSuffix]
+	if !hadDNAT && !hadSNAT {
+		return nil
+	}
+
 	var merr *multierror.Error
 	var needsFlush bool
 
@@ -1822,9 +1836,16 @@ func (r *router) DeleteDNATRule(rule firewall.Rule) error {
 		}
 	}
 
+	// Release the refcount only once the rules are gone from the kernel. On
+	// failure (including the refreshRulesMap error above) the rules and their
+	// map entries remain, keeping forwarding on until a retry removes them.
 	if merr == nil {
 		delete(r.rules, ruleKey+dnatSuffix)
 		delete(r.rules, ruleKey+snatSuffix)
+
+		if err := r.ipFwdState.ReleaseForwarding(r.af.tableFamily == nftables.TableFamilyIPv6); err != nil {
+			log.Errorf("%v", err)
+		}
 	}
 
 	return nberrors.FormatErrorOrNil(merr)

client/firewall/uspfilter/filter_filter_test.go

@@ -5,7 +5,7 @@ import (
 	"net/netip"
 	"testing"
 
-	"github.com/golang/mock/gomock"
+	"go.uber.org/mock/gomock"
 	"github.com/google/gopacket"
 	"github.com/google/gopacket/layers"
 	"github.com/stretchr/testify/require"

client/firewall/uspfilter/filter_routeacl_test.go

@@ -4,7 +4,7 @@ import (
 	"net/netip"
 	"testing"
 
-	"github.com/golang/mock/gomock"
+	"go.uber.org/mock/gomock"
 	"github.com/google/gopacket/layers"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"

client/grpc/dialer_generic.go

@@ -16,28 +16,52 @@ import (
 	"google.golang.org/grpc"
 
 	nbnet "github.com/netbirdio/netbird/client/net"
+	"github.com/netbirdio/netbird/client/netevents/sweep"
 )
 
+// Sweeper registers in-flight dials for the network change sweep.
+type Sweeper interface {
+	StartDial(ctx context.Context) *sweep.Dial
+}
+
 func WithCustomDialer(_ bool, _ string) grpc.DialOption {
+	return grpc.WithContextDialer(dialContext)
+}
+
+// WithSweeper dials like WithCustomDialer but registers connections and
+// dials with the sweeper. Append it after WithCustomDialer: gRPC applies
+// dial options in order, so the later context dialer wins.
+func WithSweeper(sweeper Sweeper) grpc.DialOption {
 	return grpc.WithContextDialer(func(ctx context.Context, addr string) (net.Conn, error) {
-		if runtime.GOOS == "linux" {
-			currentUser, err := user.Current()
-			if err != nil {
-				return nil, status.Errorf(codes.FailedPrecondition, "failed to get current user: %v", err)
-			}
+		dial := sweeper.StartDial(ctx)
+		defer dial.Release()
 
-			// the custom dialer requires root permissions which are not required for use cases run as non-root
-			if currentUser.Uid != "0" {
-				log.Debug("Not running as root, using standard dialer")
-				dialer := &net.Dialer{}
-				return dialer.DialContext(ctx, "tcp", addr)
-			}
-		}
-
-		conn, err := nbnet.NewDialer().DialContext(ctx, "tcp", addr)
+		conn, err := dialContext(dial.Ctx(), addr)
 		if err != nil {
-			return nil, fmt.Errorf("nbnet.NewDialer().DialContext: %w", err)
+			return nil, err
 		}
-		return conn, nil
+		return dial.WrapConn(conn)
 	})
 }
+
+func dialContext(ctx context.Context, addr string) (net.Conn, error) {
+	if runtime.GOOS == "linux" {
+		currentUser, err := user.Current()
+		if err != nil {
+			return nil, status.Errorf(codes.FailedPrecondition, "failed to get current user: %v", err)
+		}
+
+		// the custom dialer requires root permissions which are not required for use cases run as non-root
+		if currentUser.Uid != "0" {
+			log.Debug("Not running as root, using standard dialer")
+			dialer := &net.Dialer{}
+			return dialer.DialContext(ctx, "tcp", addr)
+		}
+	}
+
+	conn, err := nbnet.NewDialer().DialContext(ctx, "tcp", addr)
+	if err != nil {
+		return nil, fmt.Errorf("nbnet.NewDialer().DialContext: %w", err)
+	}
+	return conn, nil
+}

client/grpc/dialer_js.go

@@ -1,13 +1,26 @@
 package grpc
 
 import (
+	"context"
+
 	"google.golang.org/grpc"
 
+	"github.com/netbirdio/netbird/client/netevents/sweep"
 	"github.com/netbirdio/netbird/util/wsproxy/client"
 )
 
+// Sweeper registers in-flight dials for the network change sweep.
+type Sweeper interface {
+	StartDial(ctx context.Context) *sweep.Dial
+}
+
 // WithCustomDialer returns a gRPC dial option that uses WebSocket transport for WASM/JS environments.
 // The component parameter specifies the WebSocket proxy component path (e.g., "/management", "/signal").
 func WithCustomDialer(tlsEnabled bool, component string) grpc.DialOption {
 	return client.WithWebSocketDialer(tlsEnabled, component)
 }
+
+// WithSweeper is a no-op on WASM/JS: there is no network change signal.
+func WithSweeper(_ Sweeper) grpc.DialOption {
+	return grpc.EmptyDialOption{}
+}

client/grpc/retry.go

@@ -0,0 +1,56 @@
+package grpc
+
+import (
+	"context"
+	"errors"
+	"time"
+
+	"github.com/cenkalti/backoff/v4"
+)
+
+// ChangeWatcher exposes OS network availability transitions.
+type ChangeWatcher interface {
+	Changed() <-chan struct{}
+}
+
+// Retry mirrors backoff.Retry, but the sleep between attempts also wakes on
+// OS network availability transitions: an operation cut down by a network
+// change retries the moment the network settles instead of sleeping through
+// the recovery. A nil watcher never fires, leaving plain backoff.Retry
+// behavior.
+func Retry(ctx context.Context, operation backoff.Operation, bo backoff.BackOff, watcher ChangeWatcher) error {
+	bo.Reset()
+	for {
+		err := operation()
+		if err == nil {
+			return nil
+		}
+
+		var permanent *backoff.PermanentError
+		if errors.As(err, &permanent) {
+			return permanent.Err
+		}
+
+		next := bo.NextBackOff()
+		if next == backoff.Stop {
+			if cerr := ctx.Err(); cerr != nil {
+				return cerr
+			}
+			return err
+		}
+
+		var changed <-chan struct{}
+		if watcher != nil {
+			changed = watcher.Changed()
+		}
+		timer := time.NewTimer(next)
+		select {
+		case <-timer.C:
+		case <-changed:
+			timer.Stop()
+		case <-ctx.Done():
+			timer.Stop()
+			return ctx.Err()
+		}
+	}
+}

client/grpc/retry_test.go

@@ -0,0 +1,91 @@
+package grpc
+
+import (
+	"context"
+	"errors"
+	"testing"
+	"time"
+
+	"github.com/cenkalti/backoff/v4"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+
+	"github.com/netbirdio/netbird/client/netevents/netstate"
+)
+
+func TestRetryWakesOnNetworkChange(t *testing.T) {
+	ns := netstate.New()
+	attempts := 0
+	operation := func() error {
+		attempts++
+		if attempts == 1 {
+			return errors.New("cut by network change")
+		}
+		return nil
+	}
+
+	go func() {
+		time.Sleep(20 * time.Millisecond)
+		ns.Set(false)
+	}()
+
+	start := time.Now()
+	err := Retry(context.Background(), operation, backoff.NewConstantBackOff(time.Minute), ns)
+
+	require.NoError(t, err)
+	assert.Equal(t, 2, attempts, "network change must cause one immediate retry")
+	assert.Less(t, time.Since(start), time.Second, "the transition must cut the minute-long sleep short")
+}
+
+func TestRetryPermanentError(t *testing.T) {
+	sentinel := errors.New("permission denied")
+	operation := func() error {
+		return backoff.Permanent(sentinel)
+	}
+
+	err := Retry(context.Background(), operation, backoff.NewConstantBackOff(time.Millisecond), nil)
+	assert.ErrorIs(t, err, sentinel, "permanent errors must stop retries")
+}
+
+func TestRetryNilNetState(t *testing.T) {
+	attempts := 0
+	operation := func() error {
+		attempts++
+		if attempts < 3 {
+			return errors.New("transient")
+		}
+		return nil
+	}
+
+	err := Retry(context.Background(), operation, backoff.NewConstantBackOff(time.Millisecond), nil)
+	require.NoError(t, err)
+	assert.Equal(t, 3, attempts, "nil network state must preserve timed retries")
+}
+
+func TestRetryStops(t *testing.T) {
+	failure := errors.New("still failing")
+	operation := func() error {
+		return failure
+	}
+
+	err := Retry(context.Background(), operation, &backoff.StopBackOff{}, nil)
+	assert.ErrorIs(t, err, failure, "stop backoff must return the operation error")
+}
+
+func TestRetryCtxCancelDuringSleep(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	operation := func() error {
+		return errors.New("failing")
+	}
+
+	go func() {
+		time.Sleep(20 * time.Millisecond)
+		cancel()
+	}()
+
+	start := time.Now()
+	err := Retry(ctx, operation, backoff.NewConstantBackOff(time.Minute), netstate.New())
+
+	assert.ErrorIs(t, err, context.Canceled, "context cancellation must stop the retry loop")
+	assert.Less(t, time.Since(start), time.Second, "context cancellation must interrupt backoff sleep")
+}

client/iface/bind/ice_bind.go

@@ -22,6 +22,16 @@ import (
 	nbnet "github.com/netbirdio/netbird/client/net"
 )
 
+const (
+	// wgMsgTypeHandshakeInitiation is the lowest WireGuard message type.
+	wgMsgTypeHandshakeInitiation uint32 = 1
+	// wgMsgTypeTransport is the highest WireGuard message type.
+	wgMsgTypeTransport uint32 = 4
+	// wgMinMsgSize is the smallest WireGuard message: transport data with an empty
+	// payload, which is what a keepalive is.
+	wgMinMsgSize = 32
+)
+
 type receiverCreator struct {
 	iceBind *ICEBind
 }
@@ -47,10 +57,13 @@ type ICEBind struct {
 	endpoints   map[netip.Addr]net.Conn
 	endpointsMu sync.Mutex
 	recvChan    chan recvMessage
-	// every time when Close() is called (i.e. BindUpdate()) we need to close exit from the receiveRelayed and create a
-	// new closed channel. With the closedChanMu we can safely close the channel and create a new one
+	// Close() (i.e. BindUpdate()) closes closedChan to release receiveRelayed,
+	// and the following Open() installs a fresh one. closedChanMu guards both
+	// closedChan and closed: readers only ever hold it long enough to copy the
+	// channel, never across a blocking receive, so Open cannot be starved by a
+	// parked receiver.
 	closedChan       chan struct{}
-	closedChanMu     sync.RWMutex // protect the closeChan recreation from reading from it.
+	closedChanMu     sync.RWMutex
 	closed           bool
 	activityRecorder *ActivityRecorder
 
@@ -82,24 +95,41 @@ func NewICEBind(transportNet transport.Net, address wgaddr.Address, mtu uint16)
 }
 
 func (s *ICEBind) Open(uport uint16) ([]wgConn.ReceiveFunc, uint16, error) {
-	s.closed = false
 	s.closedChanMu.Lock()
-	s.closedChan = make(chan struct{})
-	s.closedChanMu.Unlock()
+	defer s.closedChanMu.Unlock()
+
+	// Open the underlying bind before touching any state, so a failure leaves
+	// the current generation exactly as it was. Publishing the new generation
+	// first would strand it: StdNetBind rejects an Open while it is already
+	// open, and a Close arriving in that window would mark the bind closed
+	// while this call went on to install live sockets, after which every later
+	// Close returns early and never shuts them down.
 	fns, port, err := s.StdNetBind.Open(uport)
 	if err != nil {
 		return nil, 0, err
 	}
+
+	// Release whoever is parked on the outgoing generation before replacing it.
+	// An Open that follows an Open rather than a Close would otherwise leave
+	// them waiting on a channel no later Close can reach.
+	if !s.closed {
+		close(s.closedChan)
+	}
+	s.closed = false
+	s.closedChan = make(chan struct{})
+
 	fns = append(fns, s.receiveRelayed)
 	return fns, port, nil
 }
 
 func (s *ICEBind) Close() error {
+	s.closedChanMu.Lock()
+	defer s.closedChanMu.Unlock()
+
 	if s.closed {
 		return nil
 	}
 	s.closed = true
-
 	close(s.closedChan)
 
 	s.muUDPMux.Lock()
@@ -111,6 +141,15 @@ func (s *ICEBind) Close() error {
 	return s.StdNetBind.Close()
 }
 
+// currentClosedChan copies the channel that signals the current Open
+// generation is closing. Callers select on the copy so the lock is never held
+// across a blocking receive, which would otherwise stall the next Open.
+func (s *ICEBind) currentClosedChan() chan struct{} {
+	s.closedChanMu.RLock()
+	defer s.closedChanMu.RUnlock()
+	return s.closedChan
+}
+
 func (s *ICEBind) ActivityRecorder() *ActivityRecorder {
 	return s.activityRecorder
 }
@@ -140,8 +179,10 @@ func (b *ICEBind) RemoveEndpoint(fakeIP netip.Addr) {
 }
 
 func (b *ICEBind) ReceiveFromEndpoint(ctx context.Context, ep *Endpoint, buf []byte) {
+	closedChan := b.currentClosedChan()
+
 	select {
-	case <-b.closedChan:
+	case <-closedChan:
 		return
 	case <-ctx.Done():
 		return
@@ -216,8 +257,15 @@ func (s *ICEBind) createReceiverFn(pc wgConn.BatchReader, conn *net.UDPConn, rxO
 		for i := 0; i < numMsgs; i++ {
 			msg := &(*msgs)[i]
 
-			// todo: handle err
-			if ok, _ := s.filterOutStunMessages(msg.Buffers, msg.N, msg.Addr); ok {
+			if ok, err := s.filterOutStunMessages(msg.Buffers, msg.N, msg.Addr); ok {
+				if err != nil {
+					log.Debugf("failed to handle STUN packet from %s: %v", msg.Addr, err)
+				}
+				// WireGuard reuses sizes and eps across reads and only skips a slot
+				// whose size is below the minimum message size. Leaving a consumed
+				// slot untouched makes it process this buffer again under the
+				// previous packet's length and endpoint.
+				sizes[i] = 0
 				continue
 			}
 			sizes[i] = msg.N
@@ -271,11 +319,16 @@ func (s *ICEBind) createOrUpdateMux() {
 
 func (s *ICEBind) filterOutStunMessages(buffers [][]byte, n int, addr net.Addr) (bool, error) {
 	for i := range buffers {
-		if !stun.IsMessage(buffers[i]) {
+		if n > len(buffers[i]) {
+			continue
+		}
+		pkt := buffers[i][:n]
+
+		if isWireGuardMsg(pkt) || !stun.IsMessage(pkt) {
 			continue
 		}
 
-		msg, err := s.parseSTUNMessage(buffers[i][:n])
+		msg, err := s.parseSTUNMessage(pkt)
 		if err != nil {
 			buffers[i] = []byte{}
 			return true, err
@@ -311,11 +364,10 @@ func (s *ICEBind) parseSTUNMessage(raw []byte) (*stun.Message, error) {
 // receiveRelayed is a receive function that is used to receive packets from the relayed connection and forward to the
 // WireGuard. Critical part is do not block if the Closed() has been called.
 func (c *ICEBind) receiveRelayed(buffs [][]byte, sizes []int, eps []wgConn.Endpoint) (int, error) {
-	c.closedChanMu.RLock()
-	defer c.closedChanMu.RUnlock()
+	closedChan := c.currentClosedChan()
 
 	select {
-	case <-c.closedChan:
+	case <-closedChan:
 		return 0, net.ErrClosed
 	case msg, ok := <-c.recvChan:
 		if !ok {
@@ -347,18 +399,34 @@ func putMessages(msgs *[]ipv6.Message, msgsPool *sync.Pool) {
 	msgsPool.Put(msgs)
 }
 
-func isTransportPkg(buffers [][]byte, n int) bool {
-	// The first buffer should contain at least 4 bytes for type
-	if len(buffers[0]) < 4 {
-		return true
+// isWireGuardMsg reports whether the packet carries a WireGuard message header: a
+// little-endian uint32 message type in the range 1..4, which leaves the three bytes
+// after the type byte zero, in a packet long enough to hold any WireGuard message.
+//
+// A well formed STUN message cannot take that shape. Its length field sits in the two
+// bytes the type must leave zero, and for a message of at least wgMinMsgSize bytes that
+// field holds at least 12, so the two framings do not overlap. The test has to be this
+// tight because stun.IsMessage only looks at the magic cookie, which in a WireGuard
+// message overlaps the receiver index: a session whose index happens to equal the cookie
+// would otherwise have all of its inbound data misrouted to the STUN handler until the
+// next rekey.
+func isWireGuardMsg(pkt []byte) bool {
+	if len(pkt) < wgMinMsgSize {
+		return false
 	}
 
-	// WireGuard packet type is a little-endian uint32 at start
-	packetType := binary.LittleEndian.Uint32(buffers[0][:4])
-
-	// Check if packetType matches known WireGuard message types
-	if packetType == 4 && n > 32 {
-		return true
-	}
-	return false
+	msgType := binary.LittleEndian.Uint32(pkt[:4])
+	return msgType >= wgMsgTypeHandshakeInitiation && msgType <= wgMsgTypeTransport
+}
+
+// isTransportPkg reports whether the packet is WireGuard transport data carrying a
+// payload, which is what counts as peer activity. A keepalive holds no payload and is
+// exactly wgMinMsgSize bytes.
+func isTransportPkg(buffers [][]byte, n int) bool {
+	if n < 4 || n > len(buffers[0]) {
+		return false
+	}
+
+	msgType := binary.LittleEndian.Uint32(buffers[0][:4])
+	return msgType == wgMsgTypeTransport && n > wgMinMsgSize
 }

client/iface/bind/ice_bind_close_test.go

@@ -0,0 +1,261 @@
+package bind
+
+import (
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
+	wgConn "golang.zx2c4.com/wireguard/conn"
+)
+
+// startReceivers runs every receive function the way wireguard-go's device
+// does: one goroutine per function, all tracked by a single WaitGroup. After
+// calling Bind.Close, closeBindLocked waits on exactly that WaitGroup while
+// holding device.net, so a receive function that never returns wedges the
+// device and every goroutine that needs the same lock.
+func startReceivers(fns []wgConn.ReceiveFunc) *sync.WaitGroup {
+	wg, _ := startReceiversEntered(fns)
+	return wg
+}
+
+// startReceiversEntered also returns a channel closed once every receive
+// function has been called at least once. A receive function that has been
+// entered is either inside its blocking receive or about to be, which is a
+// stronger signal to synchronise on than a bare sleep.
+func startReceiversEntered(fns []wgConn.ReceiveFunc) (*sync.WaitGroup, <-chan struct{}) {
+	var wg sync.WaitGroup
+	var entered sync.WaitGroup
+	wg.Add(len(fns))
+	entered.Add(len(fns))
+
+	for i := range fns {
+		go func(fn wgConn.ReceiveFunc) {
+			defer wg.Done()
+			buffs := [][]byte{make([]byte, 1500)}
+			sizes := make([]int, 1)
+			eps := make([]wgConn.Endpoint, 1)
+			first := true
+			for {
+				if first {
+					entered.Done()
+					first = false
+				}
+				if _, err := fn(buffs, sizes, eps); err != nil {
+					return
+				}
+			}
+		}(fns[i])
+	}
+
+	allEntered := make(chan struct{})
+	go func() {
+		entered.Wait()
+		close(allEntered)
+	}()
+	return &wg, allEntered
+}
+
+// closeBounded runs Close off the caller's goroutine so a regression that
+// wedges it fails the test instead of hanging teardown, and reports whether it
+// returned in time.
+func closeBounded(iceBind *ICEBind, timeout time.Duration) bool {
+	done := make(chan struct{})
+	go func() {
+		_ = iceBind.Close()
+		close(done)
+	}()
+	select {
+	case <-done:
+		return true
+	case <-time.After(timeout):
+		return false
+	}
+}
+
+// isClosed reports whether the bind's current generation channel is closed.
+func isClosed(iceBind *ICEBind) bool {
+	iceBind.closedChanMu.RLock()
+	ch := iceBind.closedChan
+	iceBind.closedChanMu.RUnlock()
+	select {
+	case <-ch:
+		return true
+	default:
+		return false
+	}
+}
+
+// receiversStopped reports whether every receive function returned before the
+// timeout, mirroring device.net.stopping.Wait() inside closeBindLocked.
+func receiversStopped(wg *sync.WaitGroup, timeout time.Duration) bool {
+	done := make(chan struct{})
+	go func() {
+		wg.Wait()
+		close(done)
+	}()
+	select {
+	case <-done:
+		return true
+	case <-time.After(timeout):
+		return false
+	}
+}
+
+// TestICEBindCloseReleasesReceivers covers the contract closeBindLocked relies
+// on: once Close returns, every receive function handed out by Open must stop.
+func TestICEBindCloseReleasesReceivers(t *testing.T) {
+	iceBind := setupICEBind(t)
+
+	fns, _, err := iceBind.Open(0)
+	require.NoError(t, err, "opening the bind must succeed")
+
+	wg := startReceivers(fns)
+	require.NoError(t, iceBind.Close())
+
+	require.True(t, receiversStopped(wg, 5*time.Second),
+		"every receive function must return once Close returns")
+}
+
+// TestICEBindOpenDoesNotBlockOnParkedReceiver reproduces the deadlock that
+// wedges interface creation.
+//
+// receiveRelayed used to hold closedChanMu for the whole of its blocking
+// select, so a parked receiver kept the read lock indefinitely. Open takes the
+// same mutex for writing to install a fresh closedChan, so it could never
+// acquire it while a receiver was parked. wireguard-go reaches Open from
+// Device.IpcSet and Device.Up with device.net held, so the stall takes the
+// device's lock with it and every other device goroutine queues behind it.
+//
+// Failing here means Open never returned.
+func TestICEBindOpenDoesNotBlockOnParkedReceiver(t *testing.T) {
+	iceBind := setupICEBind(t)
+
+	fns, _, err := iceBind.Open(0)
+	require.NoError(t, err, "the first Open must succeed")
+
+	wg, entered := startReceiversEntered(fns)
+	t.Cleanup(func() {
+		// Both bounded: a regression that wedges Close must surface as the
+		// assertion below, not as a hung teardown.
+		if !closeBounded(iceBind, 5*time.Second) {
+			t.Error("Close did not return during teardown; the bind lifecycle is wedged even though the assertion above passed")
+		}
+		if !receiversStopped(wg, 5*time.Second) {
+			t.Error("receive functions were still running after teardown Close, which is what closeBindLocked blocks on")
+		}
+	})
+
+	select {
+	case <-entered:
+	case <-time.After(5 * time.Second):
+		t.Fatal("receive functions never started")
+	}
+	// Entered is not yet parked, so still allow the blocking receive to be
+	// reached. Parking takes microseconds; this margin is six orders larger.
+	time.Sleep(500 * time.Millisecond)
+
+	reopened := make(chan struct{})
+	go func() {
+		// The error is irrelevant; StdNetBind rejects a second Open. What
+		// matters is that the call returns at all.
+		_, _, _ = iceBind.Open(0)
+		close(reopened)
+	}()
+
+	select {
+	case <-reopened:
+	case <-time.After(10 * time.Second):
+		t.Fatal("Open blocked while a receive function was parked; wireguard-go makes this call with device.net held, which is what stalls interface creation")
+	}
+}
+
+// TestICEBindConcurrentOpenClose exercises Open and Close from separate
+// goroutines, the way Device.IpcSet and Device.Up reach the bind, and is meant
+// to be run under -race.
+//
+// closed and closedChan must be updated together. When they were not, Close
+// could observe a stale closed and either skip close(closedChan) and
+// StdNetBind.Close entirely, leaving the receive functions running, or race a
+// second Close and close the same channel twice.
+//
+// Receive functions are deliberately not started here: this test is about the
+// shared state, and parking them would turn a race report into a hang.
+func TestICEBindConcurrentOpenClose(t *testing.T) {
+	iceBind := setupICEBind(t)
+
+	var wg sync.WaitGroup
+	wg.Add(2)
+
+	// Release both loops together so the calls genuinely interleave rather
+	// than depending on goroutine start order.
+	start := make(chan struct{})
+
+	go func() {
+		defer wg.Done()
+		<-start
+		for i := 0; i < 200; i++ {
+			_, _, _ = iceBind.Open(0)
+		}
+	}()
+	go func() {
+		defer wg.Done()
+		<-start
+		for i := 0; i < 200; i++ {
+			_ = iceBind.Close()
+		}
+	}()
+
+	close(start)
+	wg.Wait()
+	require.NoError(t, iceBind.Close())
+	require.True(t, isClosed(iceBind), "the final Close must leave the current generation channel closed")
+}
+
+// TestICEBindCloseReleasesReceiversUnderConcurrentClose runs full Open, receive,
+// Close cycles with a second Close and an Open racing the first Close. Any
+// iteration where the receive functions outlive Close, or where the surviving
+// generation channel is left open, is the state closeBindLocked deadlocks on.
+func TestICEBindCloseReleasesReceiversUnderConcurrentClose(t *testing.T) {
+	if testing.Short() {
+		t.Skip("stress test")
+	}
+
+	for i := 0; i < 200; i++ {
+		iceBind := setupICEBind(t)
+
+		fns, _, err := iceBind.Open(0)
+		require.NoError(t, err, "iteration %d: opening the bind must succeed", i)
+		wg := startReceivers(fns)
+
+		start := make(chan struct{})
+		var racers sync.WaitGroup
+		racers.Add(3)
+		for c := 0; c < 2; c++ {
+			go func() {
+				defer racers.Done()
+				<-start
+				_ = iceBind.Close()
+			}()
+		}
+		// An Open overlapping the Closes is what produces a generation whose
+		// channel outlives the flag saying the bind is closed.
+		go func() {
+			defer racers.Done()
+			<-start
+			_, _, _ = iceBind.Open(0)
+		}()
+		close(start)
+		racers.Wait()
+
+		// Settle on a closed bind whatever order the racers landed in.
+		_ = iceBind.Close()
+
+		if !receiversStopped(wg, 5*time.Second) {
+			t.Fatalf("iteration %d: receive functions still running after Close; closeBindLocked would block here on device.net.stopping.Wait", i)
+		}
+		if !isClosed(iceBind) {
+			t.Fatalf("iteration %d: Close returned with the current generation channel still open, so nothing will ever release its receivers", i)
+		}
+	}
+}

client/iface/bind/stun_filter_test.go

@@ -0,0 +1,215 @@
+//go:build !js
+
+package bind
+
+import (
+	"encoding/binary"
+	"net"
+	"testing"
+	"time"
+
+	"github.com/pion/stun/v3"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.org/x/net/ipv4"
+	wgConn "golang.zx2c4.com/wireguard/conn"
+)
+
+// magicCookieBytes is the STUN magic cookie as it appears on the wire. In a
+// WireGuard message the same offset holds the receiver (or sender) index, which is
+// a random uint32, so a session can draw exactly this value.
+var magicCookieBytes = []byte{0x21, 0x12, 0xA4, 0x42}
+
+const testBufSize = 1500
+
+// wgMsg builds a WireGuard message of the given type and size, with the index field
+// at bytes 4:8 set to index.
+func wgMsg(msgType uint32, size int, index []byte) []byte {
+	pkt := make([]byte, size)
+	binary.LittleEndian.PutUint32(pkt[:4], msgType)
+	copy(pkt[4:8], index)
+	return pkt
+}
+
+// intoBuffer copies pkt into a full-size receive buffer, the way the kernel read
+// does, so tests see the same buffer/length split as the hot path.
+func intoBuffer(pkt []byte) [][]byte {
+	buf := make([]byte, testBufSize)
+	copy(buf, pkt)
+	return [][]byte{buf}
+}
+
+func TestFilterOutStunMessages_PassesWireGuardWithCookieShapedIndex(t *testing.T) {
+	tests := []struct {
+		name    string
+		msgType uint32
+		size    int
+	}{
+		{"transport data", wgMsgTypeTransport, 128},
+		{"keepalive", wgMsgTypeTransport, wgMinMsgSize},
+		{"handshake initiation", wgMsgTypeHandshakeInitiation, 148},
+		{"handshake response", 2, 92},
+		{"cookie reply", 3, 64},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			pkt := wgMsg(tc.msgType, tc.size, magicCookieBytes)
+			require.True(t, stun.IsMessage(pkt), "precondition: pion sees this as STUN")
+
+			buffers := intoBuffer(pkt)
+			bind := &ICEBind{}
+
+			filtered, err := bind.filterOutStunMessages(buffers, tc.size, &net.UDPAddr{})
+			assert.NoError(t, err)
+			assert.False(t, filtered, "WireGuard message must be handed to WireGuard, not the STUN handler")
+			assert.Len(t, buffers[0], testBufSize, "buffer must be left intact for WireGuard")
+		})
+	}
+}
+
+func TestFilterOutStunMessages_FiltersRealSTUNMessage(t *testing.T) {
+	msg, err := stun.Build(stun.BindingRequest, stun.TransactionID, stun.Fingerprint)
+	require.NoError(t, err)
+
+	buffers := intoBuffer(msg.Raw)
+	bind := &ICEBind{}
+
+	filtered, err := bind.filterOutStunMessages(buffers, len(msg.Raw), &net.UDPAddr{})
+	assert.NoError(t, err)
+	assert.True(t, filtered, "STUN message must be consumed by the STUN handler")
+	assert.Empty(t, buffers[0], "consumed buffer must be emptied so WireGuard does not see it")
+}
+
+// TestIsWireGuardMsg_DisjointFromSTUN locks the invariant the filter relies on: a
+// well formed STUN message long enough to be a WireGuard message always has a
+// non-zero length field, so it cannot be mistaken for a WireGuard header.
+func TestIsWireGuardMsg_DisjointFromSTUN(t *testing.T) {
+	types := []stun.MessageType{
+		stun.BindingRequest,
+		stun.BindingSuccess,
+		stun.BindingError,
+		{Method: stun.MethodBinding, Class: stun.ClassIndication},
+	}
+
+	for _, msgType := range types {
+		// Long enough that the length guard is not what makes this pass.
+		msg, err := stun.Build(msgType, stun.TransactionID,
+			stun.NewUsername("remoteUfrag:localUfrag"), stun.Fingerprint)
+		require.NoError(t, err)
+		require.GreaterOrEqual(t, len(msg.Raw), wgMinMsgSize, "precondition: %s", msgType)
+		assert.False(t, isWireGuardMsg(msg.Raw),
+			"%s must not look like a WireGuard message", msgType)
+	}
+}
+
+func TestIsWireGuardMsg(t *testing.T) {
+	tests := []struct {
+		name string
+		pkt  []byte
+		want bool
+	}{
+		{"transport data", wgMsg(wgMsgTypeTransport, 128, nil), true},
+		{"handshake initiation", wgMsg(wgMsgTypeHandshakeInitiation, 148, nil), true},
+		{"unknown type 5", wgMsg(5, 128, nil), false},
+		{"type 0", wgMsg(0, 128, nil), false},
+		{"non-zero reserved byte", []byte{0x04, 0x00, 0x01, 0x00}, false},
+		{"too short", []byte{0x04, 0x00, 0x00}, false},
+		{"empty", nil, false},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			assert.Equal(t, tc.want, isWireGuardMsg(tc.pkt), "wrong classification for %s", tc.name)
+		})
+	}
+}
+
+// TestFilterOutStunMessages_IgnoresBytesBeyondPacket guards against classifying on
+// buffer contents left over from an earlier, longer packet.
+func TestFilterOutStunMessages_IgnoresBytesBeyondPacket(t *testing.T) {
+	buf := make([]byte, testBufSize)
+	copy(buf[4:8], magicCookieBytes)
+	buffers := [][]byte{buf}
+	bind := &ICEBind{}
+
+	filtered, err := bind.filterOutStunMessages(buffers, 2, &net.UDPAddr{})
+	assert.NoError(t, err)
+	assert.False(t, filtered, "a 2 byte packet must not be classified from stale buffer bytes")
+}
+
+// TestReceiveFn_ClearsSizeOfConsumedPacket covers the accounting WireGuard relies
+// on: sizes is reused across reads, so a slot whose packet was consumed as STUN must
+// be reported as empty. Otherwise WireGuard reprocesses the same buffer under the
+// previous packet's length, which for a WireGuard-shaped packet means it is handled
+// twice.
+func TestReceiveFn_ClearsSizeOfConsumedPacket(t *testing.T) {
+	conn := listenUDP(t, "udp4", "127.0.0.1:0")
+	defer conn.Close()
+
+	recvFn := receiverCreator{setupICEBind(t)}.CreateReceiverFn(
+		ipv4.NewPacketConn(conn), conn, false, createMsgPool(),
+	)
+
+	msg, err := stun.Build(stun.BindingRequest, stun.TransactionID, stun.Fingerprint)
+	require.NoError(t, err)
+
+	sender := listenUDP(t, "udp4", "127.0.0.1:0")
+	defer sender.Close()
+	_, err = sender.WriteTo(msg.Raw, conn.LocalAddr())
+	require.NoError(t, err)
+
+	require.NoError(t, conn.SetReadDeadline(time.Now().Add(3*time.Second)))
+
+	bufs := [][]byte{make([]byte, 1500)}
+	// A leftover size from an earlier read, which is what makes the missing reset
+	// observable.
+	sizes := []int{148}
+	eps := make([]wgConn.Endpoint, 1)
+
+	n, err := recvFn(bufs, sizes, eps)
+	require.NoError(t, err)
+	require.Equal(t, 1, n)
+	assert.Zero(t, sizes[0], "consumed STUN packet must not leave a size behind for WireGuard")
+}
+
+func TestIsTransportPkg(t *testing.T) {
+	tests := []struct {
+		name string
+		pkt  []byte
+		n    int
+		want bool
+	}{
+		{"transport data with payload", wgMsg(wgMsgTypeTransport, 128, nil), 128, true},
+		{"keepalive", wgMsg(wgMsgTypeTransport, wgMinMsgSize, nil), wgMinMsgSize, false},
+		{"handshake initiation", wgMsg(wgMsgTypeHandshakeInitiation, 148, nil), 148, false},
+		{"stale type bytes beyond packet", wgMsg(wgMsgTypeTransport, 128, nil), 2, false},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			assert.Equal(t, tc.want, isTransportPkg(intoBuffer(tc.pkt), tc.n),
+				"wrong activity classification for %s", tc.name)
+		})
+	}
+}
+
+// TestFilterOutStunMessages_ConsumesSTUNWithWireGuardShapedType covers the one STUN
+// encoding whose leading bytes collide with a WireGuard message type: method 0x080 as a
+// request encodes to 0x0200, so the type byte reads as a handshake response and the byte
+// after it is zero. Only the length check keeps such a message out of WireGuard's hands.
+// pion implements no method in that range, so this is a synthetic worst case rather than
+// traffic ICE produces.
+func TestFilterOutStunMessages_ConsumesSTUNWithWireGuardShapedType(t *testing.T) {
+	msg, err := stun.Build(stun.NewType(stun.Method(0x080), stun.ClassRequest), stun.TransactionID)
+	require.NoError(t, err)
+	require.Equal(t, []byte{0x02, 0x00, 0x00, 0x00}, msg.Raw[:4],
+		"precondition: the leading bytes read as a WireGuard message type")
+
+	buffers := intoBuffer(msg.Raw)
+	bind := &ICEBind{}
+
+	filtered, err := bind.filterOutStunMessages(buffers, len(msg.Raw), &net.UDPAddr{})
+	assert.NoError(t, err)
+	assert.True(t, filtered, "STUN message must be consumed despite its WireGuard-shaped type")
+}

client/iface/configurer/usp.go

@@ -502,7 +502,7 @@ func toBytes(s string) (int64, error) {
 
 func getFwmark() int {
 	if nbnet.AdvancedRouting() && runtime.GOOS == "linux" {
-		return nbnet.ControlPlaneMark
+		return int(nbnet.ControlPlaneMark)
 	}
 	return 0
 }

client/iface/device/device_filter_test.go

@@ -4,7 +4,7 @@ import (
 	"net"
 	"testing"
 
-	"github.com/golang/mock/gomock"
+	"go.uber.org/mock/gomock"
 	"github.com/google/gopacket"
 	"github.com/google/gopacket/layers"
 

client/iface/mocks/filter.go

@@ -8,7 +8,7 @@ import (
 	"net/netip"
 	reflect "reflect"
 
-	gomock "github.com/golang/mock/gomock"
+	gomock "go.uber.org/mock/gomock"
 )
 
 // MockPacketFilter is a mock of PacketFilter interface.

client/iface/mocks/tun.go

@@ -8,7 +8,7 @@ import (
 	os "os"
 	reflect "reflect"
 
-	gomock "github.com/golang/mock/gomock"
+	gomock "go.uber.org/mock/gomock"
 	tun "golang.zx2c4.com/wireguard/tun"
 )
 

client/iface/wgproxy/bind/proxy.go

@@ -53,15 +53,15 @@ func NewProxyBind(bind Bind, mtu uint16) *ProxyBind {
 	return p
 }
 
-// AddTurnConn adds a new connection to the bind.
+// AddRelayedConn adds a new connection to the bind.
 // endpoint is the NetBird address of the remote peer. The SetEndpoint return with the address what will be used in the
 // WireGuard configuration.
 //
 // Parameters:
 //   - ctx: Context is used for proxyToLocal to avoid unnecessary error messages
 //   - nbAddr: The NetBird UDP address of the remote peer, it required to generate fake address
-//   - remoteConn: The established TURN connection to the remote peer
-func (p *ProxyBind) AddTurnConn(ctx context.Context, nbAddr *net.UDPAddr, remoteConn net.Conn) error {
+//   - remoteConn: The established relayed connection to the remote peer
+func (p *ProxyBind) AddRelayedConn(ctx context.Context, nbAddr *net.UDPAddr, remoteConn net.Conn) error {
 	fakeNetIP, err := fakeAddress(nbAddr)
 	if err != nil {
 		return err

client/iface/wgproxy/ebpf/proxy.go

@@ -30,9 +30,9 @@ type WGEBPFProxy struct {
 	proxyPort         int
 	mtu               uint16
 
-	ebpfManager   ebpfMgr.Manager
-	turnConnStore map[uint16]net.Conn
-	turnConnMutex sync.Mutex
+	ebpfManager      ebpfMgr.Manager
+	relayedConnStore map[uint16]net.Conn
+	relayedConnMutex sync.Mutex
 
 	lastUsedPort uint16
 	rawConnIPv4  net.PacketConn
@@ -50,7 +50,7 @@ func NewWGEBPFProxy(wgPort int, mtu uint16) *WGEBPFProxy {
 		localWGListenPort: wgPort,
 		mtu:               mtu,
 		ebpfManager:       ebpf.GetEbpfManagerInstance(),
-		turnConnStore:     make(map[uint16]net.Conn),
+		relayedConnStore:  make(map[uint16]net.Conn),
 	}
 	return wgProxy
 }
@@ -110,14 +110,14 @@ func (p *WGEBPFProxy) Listen() error {
 	return nil
 }
 
-// AddTurnConn add new turn connection for the proxy
-func (p *WGEBPFProxy) AddTurnConn(turnConn net.Conn) (*net.UDPAddr, error) {
-	wgEndpointPort, err := p.storeTurnConn(turnConn)
+// AddRelayedConn add new relayed connection for the proxy
+func (p *WGEBPFProxy) AddRelayedConn(relayedConn net.Conn) (*net.UDPAddr, error) {
+	wgEndpointPort, err := p.storeRelayedConn(relayedConn)
 	if err != nil {
 		return nil, err
 	}
 
-	log.Infof("turn conn added to wg proxy store: %s, endpoint port: :%d", turnConn.RemoteAddr(), wgEndpointPort)
+	log.Infof("relayed conn added to wg proxy store: %s, endpoint port: :%d", relayedConn.RemoteAddr(), wgEndpointPort)
 
 	wgEndpoint := &net.UDPAddr{
 		IP:   net.ParseIP(loopbackAddr),
@@ -186,48 +186,48 @@ func (p *WGEBPFProxy) readAndForwardPacket(buf []byte) error {
 		return fmt.Errorf("failed to read UDP packet from WG: %w", err)
 	}
 
-	p.turnConnMutex.Lock()
-	conn, ok := p.turnConnStore[uint16(addr.Port)]
-	p.turnConnMutex.Unlock()
+	p.relayedConnMutex.Lock()
+	conn, ok := p.relayedConnStore[uint16(addr.Port)]
+	p.relayedConnMutex.Unlock()
 	if !ok {
 		if p.ctx.Err() == nil {
-			log.Debugf("turn conn not found by port because conn already has been closed: %d", addr.Port)
+			log.Debugf("relayed conn not found by port because conn already has been closed: %d", addr.Port)
 		}
 		return nil
 	}
 
 	if _, err := conn.Write(buf[:n]); err != nil {
-		return fmt.Errorf("failed to forward local WG packet (%d) to remote turn conn: %w", addr.Port, err)
+		return fmt.Errorf("forward local WG packet (%d) to remote relayed conn: %w", addr.Port, err)
 	}
 	return nil
 }
 
-func (p *WGEBPFProxy) storeTurnConn(turnConn net.Conn) (uint16, error) {
-	p.turnConnMutex.Lock()
-	defer p.turnConnMutex.Unlock()
+func (p *WGEBPFProxy) storeRelayedConn(relayedConn net.Conn) (uint16, error) {
+	p.relayedConnMutex.Lock()
+	defer p.relayedConnMutex.Unlock()
 
 	np, err := p.nextFreePort()
 	if err != nil {
 		return np, err
 	}
-	p.turnConnStore[np] = turnConn
+	p.relayedConnStore[np] = relayedConn
 	return np, nil
 }
 
-func (p *WGEBPFProxy) removeTurnConn(turnConnID uint16) {
-	p.turnConnMutex.Lock()
-	defer p.turnConnMutex.Unlock()
+func (p *WGEBPFProxy) removeRelayedConn(relayedConnID uint16) {
+	p.relayedConnMutex.Lock()
+	defer p.relayedConnMutex.Unlock()
 
-	_, ok := p.turnConnStore[turnConnID]
+	_, ok := p.relayedConnStore[relayedConnID]
 	if ok {
-		log.Debugf("remove turn conn from store by port: %d", turnConnID)
+		log.Debugf("remove relayed conn from store by port: %d", relayedConnID)
 	}
-	delete(p.turnConnStore, turnConnID)
+	delete(p.relayedConnStore, relayedConnID)
 }
 
 func (p *WGEBPFProxy) nextFreePort() (uint16, error) {
-	if len(p.turnConnStore) == 65535 {
-		return 0, fmt.Errorf("reached maximum turn connection numbers")
+	if len(p.relayedConnStore) == 65535 {
+		return 0, fmt.Errorf("reached maximum relayed connection numbers")
 	}
 generatePort:
 	if p.lastUsedPort == 65535 {
@@ -236,7 +236,7 @@ generatePort:
 		p.lastUsedPort++
 	}
 
-	if _, ok := p.turnConnStore[p.lastUsedPort]; ok {
+	if _, ok := p.relayedConnStore[p.lastUsedPort]; ok {
 		goto generatePort
 	}
 	return p.lastUsedPort, nil

client/iface/wgproxy/ebpf/proxy_test.go

@@ -9,32 +9,32 @@ import (
 func TestWGEBPFProxy_connStore(t *testing.T) {
 	wgProxy := NewWGEBPFProxy(1, 1280)
 
-	p, _ := wgProxy.storeTurnConn(nil)
+	p, _ := wgProxy.storeRelayedConn(nil)
 	if p != 1 {
 		t.Errorf("invalid initial port: %d", wgProxy.lastUsedPort)
 	}
 
 	numOfConns := 10
 	for i := 0; i < numOfConns; i++ {
-		p, _ = wgProxy.storeTurnConn(nil)
+		p, _ = wgProxy.storeRelayedConn(nil)
 	}
 	if p != uint16(numOfConns)+1 {
 		t.Errorf("invalid last used port: %d, expected: %d", p, numOfConns+1)
 	}
-	if len(wgProxy.turnConnStore) != numOfConns+1 {
-		t.Errorf("invalid store size: %d, expected: %d", len(wgProxy.turnConnStore), numOfConns+1)
+	if len(wgProxy.relayedConnStore) != numOfConns+1 {
+		t.Errorf("invalid store size: %d, expected: %d", len(wgProxy.relayedConnStore), numOfConns+1)
 	}
 }
 
 func TestWGEBPFProxy_portCalculation_overflow(t *testing.T) {
 	wgProxy := NewWGEBPFProxy(1, 1280)
 
-	_, _ = wgProxy.storeTurnConn(nil)
+	_, _ = wgProxy.storeRelayedConn(nil)
 	wgProxy.lastUsedPort = 65535
-	p, _ := wgProxy.storeTurnConn(nil)
+	p, _ := wgProxy.storeRelayedConn(nil)
 
-	if len(wgProxy.turnConnStore) != 2 {
-		t.Errorf("invalid store size: %d, expected: %d", len(wgProxy.turnConnStore), 2)
+	if len(wgProxy.relayedConnStore) != 2 {
+		t.Errorf("invalid store size: %d, expected: %d", len(wgProxy.relayedConnStore), 2)
 	}
 
 	if p != 2 {
@@ -46,11 +46,11 @@ func TestWGEBPFProxy_portCalculation_maxConn(t *testing.T) {
 	wgProxy := NewWGEBPFProxy(1, 1280)
 
 	for i := 0; i < 65535; i++ {
-		_, _ = wgProxy.storeTurnConn(nil)
+		_, _ = wgProxy.storeRelayedConn(nil)
 	}
 
-	_, err := wgProxy.storeTurnConn(nil)
+	_, err := wgProxy.storeRelayedConn(nil)
 	if err == nil {
-		t.Errorf("invalid turn conn store calculation")
+		t.Errorf("invalid relayed conn store calculation")
 	}
 }

client/iface/wgproxy/ebpf/wrapper.go

@@ -121,10 +121,10 @@ func NewProxyWrapper(proxy *WGEBPFProxy) *ProxyWrapper {
 	}
 }
 
-func (p *ProxyWrapper) AddTurnConn(ctx context.Context, _ *net.UDPAddr, remoteConn net.Conn) error {
-	addr, err := p.wgeBPFProxy.AddTurnConn(remoteConn)
+func (p *ProxyWrapper) AddRelayedConn(ctx context.Context, _ *net.UDPAddr, remoteConn net.Conn) error {
+	addr, err := p.wgeBPFProxy.AddRelayedConn(remoteConn)
 	if err != nil {
-		return fmt.Errorf("add turn conn: %w", err)
+		return fmt.Errorf("add relayed conn: %w", err)
 	}
 
 	headers, err := NewPacketHeaders(p.wgeBPFProxy.localWGListenPort, addr)
@@ -252,7 +252,7 @@ func (p *ProxyWrapper) CloseConn() error {
 }
 
 func (p *ProxyWrapper) proxyToLocal(ctx context.Context) {
-	defer p.wgeBPFProxy.removeTurnConn(uint16(p.wgRelayedEndpointAddr.Port))
+	defer p.wgeBPFProxy.removeRelayedConn(uint16(p.wgRelayedEndpointAddr.Port))
 
 	buf := make([]byte, p.wgeBPFProxy.mtu+bufsize.WGBufferOverhead)
 	for {
@@ -273,7 +273,7 @@ func (p *ProxyWrapper) proxyToLocal(ctx context.Context) {
 			if ctx.Err() != nil {
 				return
 			}
-			log.Errorf("failed to write out turn pkg to local conn: %v", err)
+			log.Errorf("failed to write out relayed pkg to local conn: %v", err)
 		}
 	}
 }
@@ -286,7 +286,7 @@ func (p *ProxyWrapper) readFromRemote(ctx context.Context, buf []byte) (int, err
 		}
 		p.closeListener.Notify()
 		if !errors.Is(err, io.EOF) {
-			log.Errorf("failed to read from turn conn (endpoint: :%d): %s", p.wgRelayedEndpointAddr.Port, err)
+			log.Errorf("failed to read from relayed conn (endpoint: :%d): %s", p.wgRelayedEndpointAddr.Port, err)
 		}
 		return 0, err
 	}

client/iface/wgproxy/proxy.go

@@ -7,7 +7,7 @@ import (
 
 // Proxy is a transfer layer between the relayed connection and the WireGuard
 type Proxy interface {
-	AddTurnConn(ctx context.Context, endpoint *net.UDPAddr, remoteConn net.Conn) error
+	AddRelayedConn(ctx context.Context, endpoint *net.UDPAddr, remoteConn net.Conn) error
 	EndpointAddr() *net.UDPAddr // EndpointAddr returns the address of the WireGuard peer endpoint
 	Work()                      // Work start or resume the proxy
 	Pause()                     // Pause to forward the packages from remote connection to WireGuard. The opposite way still works.

client/iface/wgproxy/proxy_test.go

@@ -95,7 +95,7 @@ func TestProxyCloseByRemoteConn(t *testing.T) {
 		t.Run(tt.name, func(t *testing.T) {
 			addr, _ := net.ResolveUDPAddr("udp", "100.108.135.221:51892")
 			relayedConn := newMockConn()
-			err := tt.proxy.AddTurnConn(ctx, addr, relayedConn)
+			err := tt.proxy.AddRelayedConn(ctx, addr, relayedConn)
 			if err != nil {
 				t.Errorf("error: %v", err)
 			}
@@ -157,7 +157,7 @@ func redirectTraffic(t *testing.T, proxy Proxy, wgPort int, endPointAddr *net.UD
 		_ = relayedServer.Close()
 	}()
 
-	if err := proxy.AddTurnConn(context.Background(), endPointAddr, relayedConn); err != nil {
+	if err := proxy.AddRelayedConn(context.Background(), endPointAddr, relayedConn); err != nil {
 		t.Errorf("error: %v", err)
 	}
 	defer func() {

client/iface/wgproxy/rawsocket/rawsocket.go

@@ -10,8 +10,6 @@ import (
 
 	log "github.com/sirupsen/logrus"
 	"golang.org/x/sys/unix"
-
-	nbnet "github.com/netbirdio/netbird/client/net"
 )
 
 // PrepareSenderRawSocketIPv4 creates and configures a raw socket for sending IPv4 packets
@@ -60,14 +58,12 @@ func prepareSenderRawSocket(family int, isIPv4 bool) (net.PacketConn, error) {
 		return nil, fmt.Errorf("binding to lo interface failed: %w", err)
 	}
 
-	// Set the fwmark on the socket.
-	err = nbnet.SetSocketOpt(fd)
-	if err != nil {
-		if closeErr := syscall.Close(fd); closeErr != nil {
-			log.Warnf("failed to close raw socket fd: %v", closeErr)
-		}
-		return nil, fmt.Errorf("setting fwmark failed: %w", err)
-	}
+	// The socket is bound to lo and only ever sends to the local WireGuard
+	// instance, a destination the local routing table resolves without help, so
+	// it carries no fwmark. Staying unmarked also keeps these packets out of
+	// third-party NAT rules that match on marks: such a rule rewriting the
+	// source would make WireGuard adopt the rewritten address as the peer
+	// endpoint.
 
 	// Convert the file descriptor to a PacketConn.
 	file := os.NewFile(uintptr(fd), fmt.Sprintf("fd %d", fd))

client/iface/wgproxy/rawsocket/rawsocket_privileged_test.go

@@ -0,0 +1,77 @@
+//go:build linux && !android && privileged
+
+package rawsocket
+
+import (
+	"net"
+	"syscall"
+	"testing"
+
+	"golang.org/x/sys/unix"
+
+	nbnet "github.com/netbirdio/netbird/client/net"
+)
+
+// The sender sockets must stay unmarked: a NAT rule matching on fwmark that
+// rewrites the source of an injected packet makes WireGuard adopt the rewritten
+// address as the peer endpoint.
+func TestSenderRawSocketsCarryNoFwmark(t *testing.T) {
+	// the mark is only ever applied when advanced routing is available, so
+	// without it the assertion below would hold for the wrong reason
+	nbnet.Init()
+	if !nbnet.AdvancedRouting() {
+		t.Skip("advanced routing unsupported, the sockets carry no mark either way")
+	}
+
+	tests := []struct {
+		name    string
+		prepare func() (net.PacketConn, error)
+		// the proxy treats the IPv6 socket as optional, so a host without IPv6
+		// is a reason to skip rather than to fail
+		optional bool
+	}{
+		{name: "IPv4", prepare: PrepareSenderRawSocketIPv4},
+		{name: "IPv6", prepare: PrepareSenderRawSocketIPv6, optional: true},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			conn, err := tc.prepare()
+			if err != nil {
+				if tc.optional {
+					t.Skipf("prepare raw socket: %v", err)
+				}
+				t.Fatalf("prepare raw socket: %v", err)
+			}
+			defer func() {
+				if err := conn.Close(); err != nil {
+					t.Logf("close raw socket: %v", err)
+				}
+			}()
+
+			syscallConn, ok := conn.(syscall.Conn)
+			if !ok {
+				t.Fatalf("raw socket %T does not expose a syscall conn", conn)
+			}
+			raw, err := syscallConn.SyscallConn()
+			if err != nil {
+				t.Fatalf("syscall conn: %v", err)
+			}
+
+			var mark int
+			var markErr error
+			if err := raw.Control(func(fd uintptr) {
+				mark, markErr = unix.GetsockoptInt(int(fd), unix.SOL_SOCKET, unix.SO_MARK)
+			}); err != nil {
+				t.Fatalf("control: %v", err)
+			}
+			if markErr != nil {
+				t.Fatalf("get SO_MARK: %v", markErr)
+			}
+
+			if mark != 0 {
+				t.Errorf("SO_MARK = %#x, want 0", mark)
+			}
+		})
+	}
+}

client/iface/wgproxy/redirect_test.go

@@ -119,9 +119,9 @@ func testRedirectAs(t *testing.T, proxy Proxy, wgPort int, nbAddr, p2pEndpoint *
 	}
 	defer relayConn.Close()
 
-	// Add TURN connection to proxy
-	if err := proxy.AddTurnConn(ctx, nbAddr, relayConn); err != nil {
-		t.Fatalf("failed to add TURN connection: %v", err)
+	// Add relayed connection to proxy
+	if err := proxy.AddRelayedConn(ctx, nbAddr, relayConn); err != nil {
+		t.Fatalf("failed to add relayed connection: %v", err)
 	}
 	defer func() {
 		if err := proxy.CloseConn(); err != nil {
@@ -304,8 +304,8 @@ func TestRedirectAs_Multiple_Switches(t *testing.T) {
 		Port: 38746,
 	}
 
-	if err := proxy.AddTurnConn(ctx, nbAddr, relayConn); err != nil {
-		t.Fatalf("failed to add TURN connection: %v", err)
+	if err := proxy.AddRelayedConn(ctx, nbAddr, relayConn); err != nil {
+		t.Fatalf("failed to add relayed connection: %v", err)
 	}
 	defer func() {
 		if err := proxy.CloseConn(); err != nil {

client/iface/wgproxy/udp/proxy.go

@@ -51,12 +51,12 @@ func NewWGUDPProxy(wgPort int, mtu uint16) *WGUDPProxy {
 	return p
 }
 
-// AddTurnConn
+// AddRelayedConn dials the local WireGuard port and stores the relayed connection.
 // The provided Context must be non-nil. If the context expires before
 // the connection is complete, an error is returned. Once successfully
 // connected, any expiration of the context will not affect the
 // connection.
-func (p *WGUDPProxy) AddTurnConn(ctx context.Context, _ *net.UDPAddr, remoteConn net.Conn) error {
+func (p *WGUDPProxy) AddRelayedConn(ctx context.Context, _ *net.UDPAddr, remoteConn net.Conn) error {
 	dialer := net.Dialer{}
 	localConn, err := dialer.DialContext(ctx, "udp", fmt.Sprintf(":%d", p.localWGListenPort))
 	if err != nil {

client/installer.nsis

@@ -22,8 +22,6 @@
 !define UI_REG_APP_PATH "Software\Microsoft\Windows\CurrentVersion\App Paths\${UI_APP_EXE}"
 !define UI_UNINSTALL_PATH "Software\Microsoft\Windows\CurrentVersion\Uninstall\${UI_APP_NAME}"
 
-!define AUTOSTART_REG_KEY "Software\Microsoft\Windows\CurrentVersion\Run"
-
 !define NETBIRD_DATA_DIR "$COMMONPROGRAMDATA\Netbird"
 
 Unicode True
@@ -228,13 +226,6 @@ WriteRegStr ${REG_ROOT} "${UNINSTALL_PATH}"  "Publisher" "${COMP_NAME}"
 
 WriteRegStr ${REG_ROOT} "${UI_REG_APP_PATH}" "" "$INSTDIR\${UI_APP_EXE}"
 
-; Autostart is owned by the UI's per-user setting (HKCU\...\Run via Wails),
-; not the installer. Drop the machine-wide entry older installers wrote so the
-; toggle is the single source of truth. HKCU is left untouched -- it may hold
-; the user's own toggle state, which must survive upgrades.
-DetailPrint "Removing installer-managed autostart registry entry if present..."
-DeleteRegValue HKLM "${AUTOSTART_REG_KEY}" "${APP_NAME}"
-
 EnVar::SetHKLM
 EnVar::AddValueEx "path" "$INSTDIR"
 
@@ -299,15 +290,6 @@ ExecWait '"$INSTDIR\${MAIN_APP_EXE}" service uninstall'
 DetailPrint "Terminating Netbird UI process..."
 ExecWait `taskkill /im ${UI_APP_EXE}.exe /f`
 
-; Remove autostart registry entries
-DetailPrint "Removing autostart registry entries if they exist..."
-; Legacy machine-wide entry written by older installers.
-DeleteRegValue HKLM "${AUTOSTART_REG_KEY}" "${APP_NAME}"
-; Per-user entry the UI toggle writes via Wails (value name is the lowercase
-; app-name slug). Uninstall removes the app, so drop it too.
-DeleteRegValue HKCU "${AUTOSTART_REG_KEY}" "${APP_NAME}"
-DeleteRegValue HKCU "${AUTOSTART_REG_KEY}" "netbird"
-
 ; Handle data deletion based on checkbox
 DetailPrint "Checking if user requested data deletion..."
 ${If} $DeleteDataEnabled == "1"

client/internal/acl/manager.go

@@ -116,11 +116,11 @@ func (d *DefaultManager) ApplyFiltering(networkMap *mgmProto.NetworkMap, dnsRout
 // firewall state, so an identical hash means an identical resulting ruleset.
 func (d *DefaultManager) firewallConfigHash(networkMap *mgmProto.NetworkMap, dnsRouteFeatureFlag bool) (uint64, error) {
 	return hashstructure.Hash(struct {
-		PeerRules            []*mgmProto.FirewallRule
-		PeerRulesIsEmpty     bool
-		RouteRules           []*mgmProto.RouteFirewallRule
-		RouteRulesIsEmpty    bool
-		DNSRouteFeatureFlag  bool
+		PeerRules           []*mgmProto.FirewallRule
+		PeerRulesIsEmpty    bool
+		RouteRules          []*mgmProto.RouteFirewallRule
+		RouteRulesIsEmpty   bool
+		DNSRouteFeatureFlag bool
 	}{
 		PeerRules:           networkMap.GetFirewallRules(),
 		PeerRulesIsEmpty:    networkMap.GetFirewallRulesIsEmpty(),
@@ -144,13 +144,13 @@ func (d *DefaultManager) applyPeerACLs(networkMap *mgmProto.NetworkMap) {
 		log.Warn("this peer is connected to a NetBird Management service with an older version. Allowing all traffic from connected peers")
 		rules = append(rules,
 			&mgmProto.FirewallRule{
-				PeerIP:    "0.0.0.0",
+				PeerIP:    "0.0.0.0", //nolint:staticcheck
 				Direction: mgmProto.RuleDirection_IN,
 				Action:    mgmProto.RuleAction_ACCEPT,
 				Protocol:  mgmProto.RuleProtocol_ALL,
 			},
 			&mgmProto.FirewallRule{
-				PeerIP:    "0.0.0.0",
+				PeerIP:    "0.0.0.0", //nolint:staticcheck
 				Direction: mgmProto.RuleDirection_OUT,
 				Action:    mgmProto.RuleAction_ACCEPT,
 				Protocol:  mgmProto.RuleProtocol_ALL,
@@ -407,7 +407,6 @@ func (d *DefaultManager) getRuleGroupingSelector(rule *mgmProto.FirewallRule) st
 	return fmt.Sprintf("%v:%v:%v:%s:%v", strconv.Itoa(int(rule.Direction)), rule.Action, rule.Protocol, rule.Port, rule.PortInfo)
 }
 
-
 // extractRuleIP extracts the peer IP from a firewall rule.
 // If sourcePrefixes is populated (new management), decode the first entry and use its address.
 // Otherwise fall back to the deprecated PeerIP string field (old management).

client/internal/acl/manager_test.go

@@ -5,9 +5,9 @@ import (
 	"net/netip"
 	"testing"
 
-	"github.com/golang/mock/gomock"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
+	"go.uber.org/mock/gomock"
 
 	"github.com/netbirdio/netbird/client/firewall"
 	"github.com/netbirdio/netbird/client/iface"
@@ -87,7 +87,7 @@ func TestDefaultManager(t *testing.T) {
 		networkMap.FirewallRules = append(
 			networkMap.FirewallRules,
 			&mgmProto.FirewallRule{
-				PeerIP:    "10.93.0.3",
+				PeerIP:    "10.93.0.3", //nolint:staticcheck
 				Direction: mgmProto.RuleDirection_IN,
 				Action:    mgmProto.RuleAction_DROP,
 				Protocol:  mgmProto.RuleProtocol_ICMP,
@@ -556,12 +556,12 @@ func TestApplyFilteringSkipsUnchangedConfig(t *testing.T) {
 
 func buildNetworkMap(peerRules, routeRules int) *mgmProto.NetworkMap {
 	nm := &mgmProto.NetworkMap{
-		FirewallRulesIsEmpty:      peerRules == 0,
+		FirewallRulesIsEmpty:       peerRules == 0,
 		RoutesFirewallRulesIsEmpty: routeRules == 0,
 	}
 	for i := range peerRules {
 		nm.FirewallRules = append(nm.FirewallRules, &mgmProto.FirewallRule{
-			PeerIP:    fmt.Sprintf("10.%d.%d.%d", i>>16&0xff, i>>8&0xff, i&0xff),
+			PeerIP:    fmt.Sprintf("10.%d.%d.%d", i>>16&0xff, i>>8&0xff, i&0xff), //nolint:staticcheck
 			Direction: mgmProto.RuleDirection_IN,
 			Action:    mgmProto.RuleAction_ACCEPT,
 			Protocol:  mgmProto.RuleProtocol_TCP,

client/internal/acl/mocks/iface_mapper.go

@@ -7,7 +7,7 @@ package mocks
 import (
 	reflect "reflect"
 
-	gomock "github.com/golang/mock/gomock"
+	gomock "go.uber.org/mock/gomock"
 	wgdevice "golang.zx2c4.com/wireguard/device"
 
 	"github.com/netbirdio/netbird/client/iface/device"

client/internal/auth/auth.go

@@ -138,26 +138,37 @@ func (a *Auth) IsSSOSupported(ctx context.Context) (bool, error) {
 
 // GetOAuthFlow returns an OAuth flow (PKCE or Device) using the existing management connection
 // This avoids creating a new connection to the management server
-func (a *Auth) GetOAuthFlow(ctx context.Context, forceDeviceAuth bool) (OAuthFlow, error) {
+func (a *Auth) GetOAuthFlow(ctx context.Context, forceDeviceAuth bool, hint string) (OAuthFlow, error) {
 	var flow OAuthFlow
-	var err error
 
-	err = a.withRetry(ctx, func(client *mgm.GrpcClient) error {
+	err := a.withRetry(ctx, func(client *mgm.GrpcClient) error {
 		if forceDeviceAuth {
-			flow, err = a.getDeviceFlow(client)
-			return err
+			deviceFlow, err := a.getDeviceFlow(client)
+			if err != nil {
+				return err
+			}
+			deviceFlow.SetLoginHint(hint)
+			flow = deviceFlow
+			return nil
 		}
 
 		// Try PKCE flow first
-		flow, err = a.getPKCEFlow(client)
+		pkceFlow, err := a.getPKCEFlow(client)
 		if err != nil {
 			// If PKCE not supported, try Device flow
 			if s, ok := status.FromError(err); ok && (s.Code() == codes.NotFound || s.Code() == codes.Unimplemented) {
-				flow, err = a.getDeviceFlow(client)
-				return err
+				deviceFlow, err := a.getDeviceFlow(client)
+				if err != nil {
+					return err
+				}
+				deviceFlow.SetLoginHint(hint)
+				flow = deviceFlow
+				return nil
 			}
 			return err
 		}
+		pkceFlow.SetLoginHint(hint)
+		flow = pkceFlow
 		return nil
 	})
 
@@ -357,6 +368,7 @@ func (a *Auth) setSystemInfoFlags(info *system.Info) {
 		a.config.EnableSSHLocalPortForwarding,
 		a.config.EnableSSHRemotePortForwarding,
 		a.config.DisableSSHAuth,
+		a.config.RemoteJobsAllowed,
 	)
 }
 

client/internal/auth/device_flow.go

@@ -304,6 +304,16 @@ func (d *DeviceAuthorizationFlow) WaitToken(ctx context.Context, info AuthFlowIn
 				return TokenInfo{}, fmt.Errorf("validate access token failed with error: %v", err)
 			}
 
+			// Same as the PKCE flow: the account the token belongs to is what
+			// callers store to send back as the login_hint. Without it a client
+			// driven through the device flow — Android TV and tvOS — never binds
+			// an account to its profile and every later login goes out blind.
+			if email, err := parseEmailFromIDToken(tokenInfo.IDToken); err != nil {
+				log.Warnf("failed to parse email from ID token: %v", err)
+			} else {
+				tokenInfo.Email = email
+			}
+
 			log.Infof("device flow: user authorization confirmed after %d polls in %s", polls, time.Since(start).Round(time.Second))
 			return tokenInfo, err
 		}

client/internal/auth/oauth.go

@@ -97,9 +97,7 @@ func authenticateWithPKCEFlow(ctx context.Context, config *profilemanager.Config
 		return nil, fmt.Errorf("getting pkce authorization flow info failed with error: %v", err)
 	}
 
-	if hint != "" {
-		pkceFlowInfo.SetLoginHint(hint)
-	}
+	pkceFlowInfo.SetLoginHint(hint)
 
 	return pkceFlowInfo, nil
 }
@@ -127,9 +125,7 @@ func authenticateWithDeviceCodeFlow(ctx context.Context, config *profilemanager.
 		}
 	}
 
-	if hint != "" {
-		deviceFlowInfo.SetLoginHint(hint)
-	}
+	deviceFlowInfo.SetLoginHint(hint)
 
 	return deviceFlowInfo, nil
 }

client/internal/conn_mgr.go

@@ -2,6 +2,7 @@ package internal
 
 import (
 	"context"
+	"maps"
 	"os"
 	"strconv"
 	"sync"
@@ -14,6 +15,7 @@ import (
 	"github.com/netbirdio/netbird/client/internal/peer"
 	"github.com/netbirdio/netbird/client/internal/peerstore"
 	"github.com/netbirdio/netbird/route"
+	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
 )
 
 // lazyForce is the resolved local decision for lazy connections, layered above the
@@ -37,11 +39,13 @@ const (
 // The only exception is ActivatePeer, which is safe for concurrent use so the
 // DNS warm-up path can call it without contending on the engine mutex.
 type ConnMgr struct {
-	peerStore        *peerstore.Store
-	statusRecorder   *peer.Status
-	iface            lazyconn.WGIface
-	force            lazyForce
-	rosenpassEnabled bool
+	peerStore      *peerstore.Store
+	statusRecorder *peer.Status
+	iface          lazyconn.WGIface
+	force          lazyForce
+	// remoteLazyEnabled caches the account-wide lazy feature flag from management.
+	// It is the default for peers that do not carry a per-peer lazy hint.
+	remoteLazyEnabled bool
 
 	lazyConnMgr *manager.Manager
 	// lazyConnMgrMu guards the lazyConnMgr pointer for readers outside the
@@ -53,6 +57,10 @@ type ConnMgr struct {
 	// (re)armed (Mode A at arm time). Injected by the engine; nil disables the reconcile.
 	reconcileRoutedIPs func(peerKey string) error
 
+	// appliedExcludeList is the exclude set last handed to the lazy manager, kept so an
+	// unchanged set on the next sync skips the O(n) reconciliation.
+	appliedExcludeList map[string]bool
+
 	wg            sync.WaitGroup
 	lazyCtx       context.Context
 	lazyCtxCancel context.CancelFunc
@@ -66,78 +74,59 @@ func (e *ConnMgr) SetRoutedIPsReconciler(fn func(peerKey string) error) {
 
 func NewConnMgr(engineConfig *EngineConfig, statusRecorder *peer.Status, peerStore *peerstore.Store, iface lazyconn.WGIface) *ConnMgr {
 	e := &ConnMgr{
-		peerStore:        peerStore,
-		statusRecorder:   statusRecorder,
-		iface:            iface,
-		force:            resolveLazyForce(engineConfig.LazyConnection),
-		rosenpassEnabled: engineConfig.RosenpassEnabled,
+		peerStore:      peerStore,
+		statusRecorder: statusRecorder,
+		iface:          iface,
+		force:          resolveLazyForce(engineConfig.LazyConnection),
 	}
 	return e
 }
 
-// Start initializes the connection manager. It starts the lazy connection manager when a
-// local override forces it on; with no local override it waits for the management feature flag.
+// Start initializes the connection manager. The lazy connection manager always runs so that
+// per-peer lazy defaults (e.g. proxy peers) work even when the account flag is off; the
+// account flag and the local override decide the default lazy state per peer (see
+// PeerLazyDefault). Rosenpass peers stay lazy-capable too: their connections just never idle
+// on their own, since rosenpass rekey traffic keeps them active.
 func (e *ConnMgr) Start(ctx context.Context) {
 	if e.lazyConnMgr != nil {
 		log.Errorf("lazy connection manager is already started")
 		return
 	}
 
-	switch e.force {
-	case lazyForceOff:
-		log.Infof("lazy connection manager is disabled by local override (%s or MDM policy)", lazyconn.EnvLazyConn)
-		e.statusRecorder.UpdateLazyConnection(false)
-		return
-	case lazyForceNone:
-		log.Infof("lazy connection manager is managed by the management feature flag")
-		e.statusRecorder.UpdateLazyConnection(false)
-		return
-	}
-
-	if e.rosenpassEnabled {
-		log.Warnf("rosenpass connection manager is enabled, lazy connection manager will not be started")
-		e.statusRecorder.UpdateLazyConnection(false)
-		return
-	}
-
 	e.initLazyManager(ctx)
-	e.statusRecorder.UpdateLazyConnection(true)
+	e.statusRecorder.UpdateLazyConnection(e.PeerLazyDefault(mgmProto.LazyState_LazyStateDefault))
 }
 
-// UpdatedRemoteFeatureFlag is called when the remote feature flag is updated.
-// If enabled, it initializes the lazy connection manager and start it. Do not need to call Start() again.
-// If disabled, then it closes the lazy connection manager and open the connections to all peers.
-func (e *ConnMgr) UpdatedRemoteFeatureFlag(ctx context.Context, enabled bool) error {
-	// a local override (NB_LAZY_CONN or local config) takes precedence over management
-	if e.force != lazyForceNone {
-		return nil
+// UpdatedRemoteFeatureFlag caches the account-wide lazy feature flag. The manager itself is
+// not started or stopped here; the per-sync exclude-list reconciliation moves normal peers
+// between the lazy and always-active sets when the flag flips.
+func (e *ConnMgr) UpdatedRemoteFeatureFlag(_ context.Context, enabled bool) error {
+	e.remoteLazyEnabled = enabled
+	if e.isStartedWithLazyMgr() {
+		e.statusRecorder.UpdateLazyConnection(e.PeerLazyDefault(mgmProto.LazyState_LazyStateDefault))
+	}
+	return nil
+}
+
+// PeerLazyDefault reports whether a peer should be lazy. The local override
+// (NB_LAZY_CONN/MDM) wins over everything; without a local override the
+// management per-peer state applies (LazyStateLazy/Eager force the decision),
+// and LazyStateDefault follows the account-wide flag.
+func (e *ConnMgr) PeerLazyDefault(state mgmProto.LazyState) bool {
+	switch e.force {
+	case lazyForceOn:
+		return true
+	case lazyForceOff:
+		return false
 	}
 
-	if enabled {
-		// if the lazy connection manager is already started, do not start it again
-		if e.lazyConnMgr != nil {
-			return nil
-		}
-
-		if e.rosenpassEnabled {
-			log.Infof("rosenpass connection manager is enabled, lazy connection manager will not be started")
-			e.statusRecorder.UpdateLazyConnection(false)
-			return nil
-		}
-
-		log.Infof("lazy connection manager is enabled by the management feature flag")
-		e.initLazyManager(ctx)
-		e.statusRecorder.UpdateLazyConnection(true)
-		return e.addPeersToLazyConnManager()
-	} else {
-		if e.lazyConnMgr == nil {
-			e.statusRecorder.UpdateLazyConnection(false)
-			return nil
-		}
-		log.Infof("lazy connection manager is disabled by management feature flag")
-		e.closeManager(ctx)
-		e.statusRecorder.UpdateLazyConnection(false)
-		return nil
+	switch state {
+	case mgmProto.LazyState_LazyStateLazy:
+		return true
+	case mgmProto.LazyState_LazyStateEager:
+		return false
+	default:
+		return e.remoteLazyEnabled
 	}
 }
 
@@ -157,6 +146,13 @@ func (e *ConnMgr) SetExcludeList(ctx context.Context, peerIDs map[string]bool) {
 		return
 	}
 
+	// The exclude set is recomputed every sync but rarely changes; skip the O(n)
+	// store lookups and reconciliation when it matches what was already applied.
+	if maps.Equal(peerIDs, e.appliedExcludeList) {
+		return
+	}
+	e.appliedExcludeList = maps.Clone(peerIDs)
+
 	excludedPeers := make([]lazyconn.PeerConfig, 0, len(peerIDs))
 
 	for peerID := range peerIDs {
@@ -192,12 +188,16 @@ func (e *ConnMgr) SetExcludeList(ctx context.Context, peerIDs map[string]bool) {
 	}
 }
 
-func (e *ConnMgr) AddPeerConn(ctx context.Context, peerKey string, conn *peer.Conn) (exists bool) {
+// AddPeerConn registers a peer connection. permanent requests an always-active connection
+// (the peer belongs to the exclude set: a forwarder, or a peer that is not lazy by policy).
+// Non-permanent peers are handed to the lazy manager. The subsequent SetExcludeList call
+// reconciles membership for existing peers across flag flips.
+func (e *ConnMgr) AddPeerConn(ctx context.Context, peerKey string, conn *peer.Conn, permanent bool) (exists bool) {
 	if success := e.peerStore.AddPeerConn(peerKey, conn); !success {
 		return true
 	}
 
-	if !e.isStartedWithLazyMgr() {
+	if !e.isStartedWithLazyMgr() || permanent {
 		if err := conn.Open(ctx); err != nil {
 			conn.Log.Errorf("failed to open connection: %v", err)
 		}
@@ -296,6 +296,8 @@ func (e *ConnMgr) Close() {
 	e.lazyConnMgrMu.Lock()
 	e.lazyConnMgr = nil
 	e.lazyConnMgrMu.Unlock()
+
+	e.appliedExcludeList = nil
 }
 
 func (e *ConnMgr) initLazyManager(engineCtx context.Context) {
@@ -309,6 +311,8 @@ func (e *ConnMgr) initLazyManager(engineCtx context.Context) {
 	e.lazyCtx, e.lazyCtxCancel = context.WithCancel(engineCtx)
 	e.lazyConnMgrMu.Unlock()
 
+	e.appliedExcludeList = nil
+
 	e.wg.Add(1)
 	go func() {
 		defer e.wg.Done()
@@ -316,46 +320,6 @@ func (e *ConnMgr) initLazyManager(engineCtx context.Context) {
 	}()
 }
 
-func (e *ConnMgr) addPeersToLazyConnManager() error {
-	peers := e.peerStore.PeersPubKey()
-	lazyPeerCfgs := make([]lazyconn.PeerConfig, 0, len(peers))
-	for _, peerID := range peers {
-		var peerConn *peer.Conn
-		var exists bool
-		if peerConn, exists = e.peerStore.PeerConn(peerID); !exists {
-			log.Warnf("failed to find peer conn for peerID: %s", peerID)
-			continue
-		}
-
-		lazyPeerCfg := lazyconn.PeerConfig{
-			PublicKey:  peerID,
-			AllowedIPs: peerConn.WgConfig().AllowedIps,
-			PeerConnID: peerConn.ConnID(),
-			Log:        peerConn.Log,
-		}
-		lazyPeerCfgs = append(lazyPeerCfgs, lazyPeerCfg)
-	}
-
-	return e.lazyConnMgr.AddActivePeers(lazyPeerCfgs)
-}
-
-func (e *ConnMgr) closeManager(ctx context.Context) {
-	if e.lazyConnMgr == nil {
-		return
-	}
-
-	e.lazyCtxCancel()
-	e.wg.Wait()
-
-	e.lazyConnMgrMu.Lock()
-	e.lazyConnMgr = nil
-	e.lazyConnMgrMu.Unlock()
-
-	for _, peerID := range e.peerStore.PeersPubKey() {
-		e.peerStore.PeerConnOpen(ctx, peerID)
-	}
-}
-
 func (e *ConnMgr) isStartedWithLazyMgr() bool {
 	return e.lazyConnMgr != nil && e.lazyCtxCancel != nil
 }