Updating changelog + build for 3.11.1

Add dummy SWer
Add default precision
2026-05-07 22:05:14 +02:00 · 2026-04-12 20:50:12 -07:00 · 2026-04-12 19:11:17 -07:00 · 2026-04-12 18:59:44 -07:00 · 2026-04-12 17:40:37 -07:00 · 2026-04-12 17:35:54 -07:00
310 changed files with 18038 additions and 8059 deletions
@@ -11,7 +11,7 @@ jobs:

    steps:
      - name: Check out repository
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6

      - name: Set up Python
        uses: actions/setup-python@v6
@@ -44,7 +44,7 @@ jobs:

    steps:
      - name: Check out repository
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6

      - name: Set up Node.js
        uses: actions/setup-node@v6
@@ -0,0 +1,73 @@
+name: Publish AUR package
+
+# Pushes the contents of pkg/aur/ to the remoteterm-meshcore AUR repository
+# whenever a GitHub release is published. Can also be triggered manually for
+# testing or out-of-band republishes.
+#
+# Required secrets:
+#   AUR_SSH_PRIVATE_KEY  Private SSH key registered with the AUR maintainer
+#                        account that owns the remoteterm-meshcore package.
+#   AUR_COMMIT_EMAIL     Email used for the AUR git commit identity.
+
+on:
+  release:
+    types: [published]
+  workflow_dispatch:
+    inputs:
+      version:
+        description: 'Version to publish (no v prefix, e.g. 3.9.1)'
+        required: true
+
+concurrency:
+  # Serialize publishes so a fast back-to-back release sequence cannot race
+  # two pushes against the AUR repo. The later one wins by virtue of being
+  # the final state.
+  group: publish-aur
+  cancel-in-progress: false
+
+jobs:
+  publish-aur:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v6
+
+      - name: Resolve version from event
+        id: version
+        run: |
+          if [ "${{ github.event_name }}" = "workflow_dispatch" ]; then
+            VERSION="${{ inputs.version }}"
+          else
+            VERSION="${{ github.event.release.tag_name }}"
+          fi
+          VERSION="${VERSION#v}"
+          echo "version=$VERSION" >> "$GITHUB_OUTPUT"
+          echo "Publishing AUR package for version $VERSION"
+
+      - name: Stamp pkgver into PKGBUILD
+        run: |
+          sed -i "s/^pkgver=.*/pkgver=${{ steps.version.outputs.version }}/" pkg/aur/PKGBUILD
+          sed -i "s/^pkgrel=.*/pkgrel=1/" pkg/aur/PKGBUILD
+
+      - name: Publish to AUR
+        uses: KSXGitHub/github-actions-deploy-aur@v4.1.2
+        with:
+          pkgname: remoteterm-meshcore
+          pkgbuild: pkg/aur/PKGBUILD
+          assets: |
+            pkg/aur/remoteterm-meshcore.install
+            pkg/aur/remoteterm-meshcore.service
+            pkg/aur/remoteterm-meshcore.sysusers
+            pkg/aur/remoteterm-meshcore.tmpfiles
+            pkg/aur/remoteterm.env
+          commit_username: jackkingsman
+          commit_email: ${{ secrets.AUR_COMMIT_EMAIL }}
+          ssh_private_key: ${{ secrets.AUR_SSH_PRIVATE_KEY }}
+          commit_message: "Update to ${{ steps.version.outputs.version }}"
+          # Recompute sha256sums from the live release tarball + the bundled
+          # service/env files. The committed PKGBUILD has SKIP placeholders.
+          updpkgsums: true
+          # Validate the PKGBUILD parses and sources download, but skip the
+          # actual build (which would run uv sync + npm install for several
+          # minutes of CI time on every release).
+          test: true
+          test_flags: --clean --cleanbuild --nodeps --nobuild
@@ -30,3 +30,6 @@ references/
 docker-compose.yml
 docker-compose.yaml
 .docker-certs/
+
+# HA test environment (created by scripts/setup/start_ha_test_env.sh)
+ha_test_config/
@@ -209,6 +209,7 @@ This message-layer echo/path handling is independent of raw-packet storage dedup
 │   │       ├── MapView.tsx       # Leaflet map showing node locations
 │   │       └── ...
 │   └── vite.config.ts
+├── pkg/aur/                # AUR package files (PKGBUILD, systemd service, env, install hooks)
 ├── scripts/                # Quality / release helpers (listing below is representative, not exhaustive)
 │   ├── build/
 │   │   ├── collect_licenses.sh # Gather third-party license attributions
@@ -216,7 +217,8 @@ This message-layer echo/path handling is independent of raw-packet storage dedup
 │   ├── quality/
 │   │   ├── all_quality.sh      # Repo-standard autofix + validate gate
 │   │   ├── e2e.sh              # End-to-end test runner
-│   │   └── extended_quality.sh # Quality gate plus e2e and Docker matrix
+│   │   ├── extended_quality.sh # Quality gate plus e2e and Docker matrix
+│   │   └── test_aur_package.sh # Build + install AUR package in Arch Docker containers
 │   └── setup/
 │       ├── fetch_prebuilt_frontend.py # Download release frontend fallback
 │       └── install_service.sh         # Install/configure Linux systemd service
@@ -327,6 +329,7 @@ All endpoints are prefixed with `/api` (e.g., `/api/health`).
 | GET | `/api/contacts/analytics` | Unified keyed-or-name contact analytics payload |
 | GET | `/api/contacts/repeaters/advert-paths` | List recent unique advert paths for all contacts |
 | POST | `/api/contacts` | Create contact (optionally trigger historical DM decrypt) |
+| POST | `/api/contacts/bulk-delete` | Delete multiple contacts |
 | DELETE | `/api/contacts/{public_key}` | Delete contact |
 | POST | `/api/contacts/{public_key}/mark-read` | Mark contact conversation as read |
 | POST | `/api/contacts/{public_key}/command` | Send CLI command to repeater |
@@ -346,12 +349,13 @@ All endpoints are prefixed with `/api` (e.g., `/api/health`).
 | POST | `/api/contacts/{public_key}/room/status` | Fetch room-server status telemetry |
 | POST | `/api/contacts/{public_key}/room/lpp-telemetry` | Fetch room-server CayenneLPP sensor data |
 | POST | `/api/contacts/{public_key}/room/acl` | Fetch room-server ACL entries |
-
 | GET | `/api/channels` | List channels |
 | GET | `/api/channels/{key}/detail` | Comprehensive channel profile (message stats, top senders) |
 | POST | `/api/channels` | Create channel |
+| POST | `/api/channels/bulk-hashtag` | Create multiple hashtag channels |
 | DELETE | `/api/channels/{key}` | Delete channel |
 | POST | `/api/channels/{key}/flood-scope-override` | Set or clear a per-channel regional flood-scope override |
+| POST | `/api/channels/{key}/path-hash-mode-override` | Set or clear a per-channel path hash mode override |
 | POST | `/api/channels/{key}/mark-read` | Mark channel as read |
 | GET | `/api/messages` | List with filters (`q`, `after`/`after_id` for forward pagination) |
 | GET | `/api/messages/around/{id}` | Get messages around a specific message (for jump-to-message) |
@@ -369,7 +373,7 @@ All endpoints are prefixed with `/api` (e.g., `/api/health`).
 | POST | `/api/settings/favorites/toggle` | Toggle favorite status |
 | POST | `/api/settings/blocked-keys/toggle` | Toggle blocked key |
 | POST | `/api/settings/blocked-names/toggle` | Toggle blocked name |
-| POST | `/api/settings/migrate` | One-time migration from frontend localStorage |
+| POST | `/api/settings/tracked-telemetry/toggle` | Toggle tracked telemetry repeater |
 | GET | `/api/fanout` | List all fanout configs |
 | POST | `/api/fanout` | Create new fanout config |
 | PATCH | `/api/fanout/{id}` | Update fanout config (triggers module reload) |
@@ -400,6 +404,7 @@ All endpoints are prefixed with `/api` (e.g., `/api/health`).
 - Hashtag channels: `SHA256("#name")[:16]` converted to hex
 - Custom channels: User-provided or generated
 - Channels may also persist `flood_scope_override`; when set, channel sends temporarily switch the radio flood scope to that value for the duration of the send, then restore the global app setting.
+- Channels may persist `path_hash_mode_override` (0/1/2); when set, channel sends temporarily switch the radio path hash mode for the duration of the send, then restore the radio default.

 ### Message Types

@@ -463,7 +468,7 @@ mc.subscribe(EventType.ACK, handler)
 |----------|---------|-------------|
 | `MESHCORE_SERIAL_PORT` | auto-detect | Serial port for radio |
 | `MESHCORE_TCP_HOST` | *(none)* | TCP host for radio (mutually exclusive with serial/BLE) |
-| `MESHCORE_TCP_PORT` | `4000` | TCP port (used with `MESHCORE_TCP_HOST`) |
+| `MESHCORE_TCP_PORT` | `5000` | TCP port (used with `MESHCORE_TCP_HOST`) |
 | `MESHCORE_BLE_ADDRESS` | *(none)* | BLE device address (mutually exclusive with serial/TCP) |
 | `MESHCORE_BLE_PIN` | *(required with BLE)* | BLE PIN code |
 | `MESHCORE_SERIAL_BAUDRATE` | `115200` | Serial baud rate |
@@ -475,7 +480,7 @@ mc.subscribe(EventType.ACK, handler)
 | `MESHCORE_ENABLE_MESSAGE_POLL_FALLBACK` | `false` | Switch the always-on radio audit task from hourly checks to aggressive 10-second polling; the audit checks both missed message drift and channel-slot cache drift |
 | `MESHCORE_FORCE_CHANNEL_SLOT_RECONFIGURE` | `false` | Disable channel-slot reuse and force `set_channel(...)` before every channel send, even on serial/BLE |

-**Note:** Runtime app settings are stored in the database (`app_settings` table), not environment variables. These include `max_radio_contacts`, `auto_decrypt_dm_on_advert`, `sidebar_sort_order`, `advert_interval`, `last_advert_time`, `favorites`, `last_message_times`, `flood_scope`, `blocked_keys`, and `blocked_names`. `max_radio_contacts` is the configured radio contact capacity baseline used by background maintenance: favorites reload first, non-favorite fill targets about 80% of that value, and full offload/reload triggers around 95% occupancy. They are configured via `GET/PATCH /api/settings`. The backend still carries `sidebar_sort_order` for compatibility and migration, but the current frontend sidebar stores sort order per section (`Channels`, `Contacts`, `Repeaters`) in localStorage rather than treating it as one shared server-backed preference. MQTT, bot, webhook, Apprise, and SQS configs are stored in the `fanout_configs` table, managed via `/api/fanout`. If the radio's channel slots appear unstable or another client is mutating them underneath this app, operators can force the old always-reconfigure send path with `MESHCORE_FORCE_CHANNEL_SLOT_RECONFIGURE=true`.
+**Note:** Runtime app settings are stored in the database (`app_settings` table), not environment variables. These include `max_radio_contacts`, `auto_decrypt_dm_on_advert`, `advert_interval`, `last_advert_time`, `last_message_times`, `flood_scope`, `blocked_keys`, `blocked_names`, `discovery_blocked_types`, `tracked_telemetry_repeaters`, and `auto_resend_channel`. `max_radio_contacts` is the configured radio contact capacity baseline used by background maintenance: favorites reload first, non-favorite fill targets about 80% of that value, and full offload/reload triggers around 95% occupancy. They are configured via `GET/PATCH /api/settings`. MQTT, bot, webhook, Apprise, and SQS configs are stored in the `fanout_configs` table, managed via `/api/fanout`. If the radio's channel slots appear unstable or another client is mutating them underneath this app, operators can force the old always-reconfigure send path with `MESHCORE_FORCE_CHANNEL_SLOT_RECONFIGURE=true`.

 Byte-perfect channel retries are user-triggered via `POST /api/messages/channel/{message_id}/resend` and are allowed for 30 seconds after the original send.

@@ -1,3 +1,87 @@
+## [3.11.1] - 2026-04-12
+
+* Feature: Home Assistant MQTT fanout
+* Feature: Add dummy service worker to enable PWA
+* Bugfix: DB connection plurality issues
+* Misc: Migration improvements
+* Misc: Search keys from beginning
+
+## [3.11.0] - 2026-04-10
+
+* Feature: Radio health and contact data accessible on fanout bus
+* Feature: Local node radio stats (voltage etc.) on WS health bus
+* Feature: Battery indicator optional in status bar (configured in Local Settings)
+* Bugfix: Fix same-second same-message collision in room servers
+* Bugfix: Don't consume DM resend attempt if the radio was just busy
+* Bugfix: Assume that a same-second same-message same-first-byte-key DM is more likely an echo than them sending the same message
+* Bugfix: Multi-retry for flood scope restoration
+* Misc: Testing & documentation improvements
+
+## [3.10.0] - 2026-04-10
+
+* Feature: Add Arch AUR package
+* Feature: 72hr packet density view in statistics
+* Feature: Add warnings for event loop selection for MQTT on Windows startup
+* Bugfix: Bump Apprise to 1.9.9 to fix Matrix bug
+* Misc: More memory-conscious on recent contact fetch
+* Misc: Fix statistics pane e2e test
+
+## [3.9.0] - 2026-04-06
+
+* Feature: Add hop counts to hop-width selection options
+* Feature: Show cached repeater telemetry inline in settings
+* Feature: Retain recent traces and make them click-to-re-run
+* Feature: Autofocus channel/DM textbox on desktop
+* Feature: Favorites on the radio are now imported as favorites
+* Bugfix: Be clearer on issue identification for missing HTTPS context in channel finder
+* Bugfix: Don't use sender timestamp for message sequence display
+* Bugfix: Function on subdomains happily
+* Misc: Be gentler, room s/cracker/finder/
+* Misc: Test and frontend correctness & test fixes
+* Misc: Don't repeat clock sync failure logs
+* Misc: Make warning in readme clearer about taking over the radio
+* Misc: Improve readme phrasings
+* Misc: Better y-axis selection for battery read-out
+* Misc: Provide clearer warning on docker setup without docker installed
+* Misc: Default visualizer stale pruning to on/5 minutes
+* Misc: Migrate favorites to better storage pattern
+* Misc: Provide dumper script for API + WS interfaces for prep for HA integration
+
+## [3.8.0] - 2026-04-03
+
+* Feature: Per-channel hop width override
+* Feature: Intervalized repeater telemetry collection
+* Feature: Auto-resend option for byte-perfect resends on no repeater echo
+* Feature: Attach RSSI/SNR to received packets
+* Feature: Add motion packet display to map
+* Feature: Map dark mode
+* Bugfix: Make DB indices more useful around capitalization
+* Misc: Bump required Python to 3.11
+* Misc: Performance, documentation, and test improvements
+* Misc: More yields during long radio operations
+* Misc: Dead code & crufty test removal
+* Misc: Remove all but stub frontend favorites migration for very very old versions
+
+## [3.7.1] - 2026-04-02
+
+* Feature: Redact Apprise URLs to prevent sensitive information disclosure
+
+## [3.7.0] - 2026-04-02
+
+* Feature: Repeater battery tracking
+* Feature: Repeater info pane just like contacts
+* Feature: Make repeaters blockable
+* Feature: Add new-node advert blocking
+* Feature: Add bulk deletion interface
+* Feature: Bulk room add on alt+click of new channel button
+* Feature: More info in debug endpoint
+* Bugfix: Be more conservative around radio load limits and don't exceed radio-reported capacity
+* Misc: Default auto-DM decrypt to true
+* Misc: Reorganize some settings panes
+* Misc: Enable FK pragma
+* Misc: Various performance and correctness fixes
+* Misc: Correct TCP default port
+
 ## [3.6.7] - 2026-03-31

 * Misc: Remove armv7 (for now)
@@ -80,7 +164,7 @@
 * Bugfix: Fix Apprise duplicate names
 * Bugfix: Be better about identity resolution in the stats pane
 * Misc: Docs, test, and performance enhancements
-* Misc: Don't prompt "Are you sure" when leaving an unedited interation
+* Misc: Don't prompt "Are you sure" when leaving an unedited integration
 * Misc: Log node time on startup
 * Misc: Improve community MQTT error bubble-up
 * Misc: Unread DMs always have a red unread counter
@@ -107,7 +191,7 @@
 ## [3.3.0] - 2026-03-13

 * Feature: Use dashed lines to show collapsed ambiguous router results
-* Feature: Jump to unred
+* Feature: Jump to unread
 * Feature: Local channel management to prevent need to reload channel every time
 * Feature: Debug endpoint
 * Feature: Force-singleton channel management
@@ -170,7 +254,7 @@
 * Feature: Massive codebase refactor and overhaul
 * Bugfix: Fix packet parsing for trace packets
 * Bugfix: Refetch channels on reconnect
-* Bugfix: Load All on repeater pane on mobile doesn't etend into lower text
+* Bugfix: Load All on repeater pane on mobile doesn't extend into lower text
 * Bugfix: Timestamps in logs
 * Bugfix: Correct wrong clock sync command
 * Misc: Improve bot error bubble up
@@ -187,10 +271,6 @@

 * Bugfix: Don't obscure new integration dropdown on session boundary

-## [2.7.8] - 2026-03-08
-
-
-
 ## [2.7.8] - 2026-03-08

 * Bugfix: Improve frontend asset resolution and fixup the build/push script
@@ -70,17 +70,111 @@ npm run test:run
 npm run build
 ```

+## Quality + Publishing Scripts
+
+<details>
+<summary>scripts/quality/</summary>
+
+| Script | Purpose |
+|--------|---------|
+| `all_quality.sh` | Repo-standard gate: autofix (ruff, eslint, prettier), then pyright, pytest, vitest, and frontend build. Run before finishing any code change. |
+| `extended_quality.sh` | `all_quality.sh` plus e2e tests and Docker build matrix. Used for release validation. |
+| `e2e.sh` | Thin wrapper that runs Playwright e2e tests from `tests/e2e/`. |
+| `docker_ci.sh` | Builds the Docker image and runs a smoke test against it. |
+| `test_aur_package.sh` | Builds the AUR package in an Arch container, then installs and boots it in a second container with port 8000 exposed (hang finish). |
+| `run_aur_with_radio.sh` | Like `test_aur_package.sh` but passes through the host serial device for testing with a real radio (hang finish). |
+
+</details>
+
+<details>
+<summary>scripts/build/</summary>
+
+| Script | Purpose |
+|--------|---------|
+| `publish.sh` | Full release ceremony: quality gate, version bump, changelog, frontend build, Docker multi-arch push, GitHub release. |
+| `release_common.sh` | Shared shell helpers (version validation, formatting) sourced by other build scripts. |
+| `package_release_artifact.sh` | Builds the prebuilt-frontend release zip attached to GitHub releases. |
+| `push_docker_multiarch.sh` | Builds and pushes multi-arch Docker images (amd64 + arm64). |
+| `create_github_release.sh` | Creates a GitHub release with changelog notes and the release artifact. |
+| `extract_release_notes.sh` | Extracts the latest version's notes from `CHANGELOG.md` for the release body. |
+| `collect_licenses.sh` | Gathers third-party license attributions into `LICENSES.md`. |
+| `print_frontend_licenses.cjs` | Helper that extracts frontend npm dependency licenses. |
+| `dump_api_specs.py` | Dumps the OpenAPI spec from the running backend (developer utility). |
+
+</details>
+
 ## E2E Testing

-E2E coverage exists, but it is intentionally not part of the normal development path.
+E2E tests exercise the full stack (backend + frontend + real radio hardware) via Playwright.

-These tests are only guaranteed to run correctly in a narrow subset of environments; they require a busy mesh with messages arriving constantly, an available autodetect-able radio, and a contact in the test database (which you can provide in `tests/e2e/.tmp/e2e-test.db` after an initial run). E2E tests are generally not necessary to run for normal development work.
+> [!WARNING]
+> E2E tests are **not part of the normal development path** — most contributors will never need to run them. They exist to catch integration issues that unit tests can't and generally only need to be run by maintainers.
+
+### Hardware requirements
+
+- A MeshCore radio connected via serial (auto-detected, or set `MESHCORE_SERIAL_PORT`)
+- The radio must be powered on and past its startup sequence before tests begin
+
+### Running

 ```bash
 cd tests/e2e
 npm install
-npx playwright test # headless
-npx playwright test --headed # you can probably guess
+npx playwright install chromium  # first time only
+npx playwright test              # headless
+npx playwright test --headed     # watch it run
+```
+
+The test harness starts its own uvicorn instance on port 8001 with a fresh temporary database. Your development server (port 8000) is unaffected.
+
+### Test tiers
+
+**Most tests (22 of 28) are fully self-contained.** They seed their own data via API calls or direct DB writes and need only a connected radio. These cover messaging, pagination, search, favorites, settings, fanout integrations, historical decryption, and all UI-only views.
+
+**Mesh-traffic tests (tagged `@mesh-traffic`)** wait up to 3 minutes for an incoming message from another node on the network. If no traffic arrives, they fail with an advisory that the failure may be RF conditions, not a bug. These are: `incoming-message` and `packet-feed` (second test only).
+
+**The partner-radio DM ACK test (tagged `@partner-radio`)** validates direct-route learning by sending a DM and waiting for an ACK. It requires a second radio in range that has your test radio in its contacts. Configure the partner node's public key and name via `E2E_PARTNER_RADIO_PUBKEY` and `E2E_PARTNER_RADIO_NAME`.
+
+### Making mesh-traffic tests reliable: the echo bot
+
+The most practical way to guarantee incoming traffic is to run an **echo bot on a second radio** monitoring a known channel. When the test suite starts a `@mesh-traffic` test, it sends a trigger message to that channel. If a bot on another radio is listening, it replies — generating the incoming RF packet the test needs within seconds instead of waiting for organic mesh traffic.
+
+The test suite sends `!echo please give incoming message` to the echo channel (default `#flightless`) at the start of each `@mesh-traffic` test. The trigger message is configurable via `E2E_ECHO_TRIGGER_MESSAGE`.
+
+Setup:
+1. Set up a second MeshCore radio within RF range of your test radio
+2. Run a RemoteTerm instance on the second radio
+3. Configure a bot on the second radio that monitors the echo channel and replies when it sees the trigger. Example bot code:
+   ```python
+   def bot(sender_name, sender_key, message_text, is_dm,
+           channel_key, channel_name, sender_timestamp, path):
+       if "!echo" in message_text.lower():
+           return f"[ECHO] {message_text}"
+       return None
+   ```
+4. The test suite calls `nudgeEchoBot()` automatically — no manual intervention needed
+
+Without the echo bot, `@mesh-traffic` tests rely on organic traffic from other nodes. In a quiet RF environment they will time out.
+
+### Environment variables
+
+All E2E environment configuration is centralized in `tests/e2e/helpers/env.ts` with defaults that work for the maintainer's test rig. Override via environment variables:
+
+| Variable | Default | Purpose |
+|----------|---------|---------|
+| `MESHCORE_SERIAL_PORT` | auto-detect | Serial port for the test radio |
+| `E2E_ECHO_CHANNEL` | `#flightless` | Channel the echo bot monitors for traffic generation |
+| `E2E_ECHO_TRIGGER_MESSAGE` | `!echo please give incoming message` | Message sent to nudge the echo bot |
+| `E2E_PARTNER_RADIO_PUBKEY` | *(maintainer's test node)* | 64-char hex public key of a node that will ACK DMs from your radio |
+| `E2E_PARTNER_RADIO_NAME` | *(maintainer's test node)* | Display name of that node (used in UI assertions) |
+
+Example for a contributor with their own two-radio setup:
+
+```bash
+E2E_ECHO_CHANNEL="#mytest" \
+E2E_PARTNER_RADIO_PUBKEY="abcd1234...full64charhexkey..." \
+E2E_PARTNER_RADIO_NAME="MyTestNode" \
+npx playwright test
 ```

 ## Pull Request Expectations
@@ -56,7 +56,7 @@ SOFTWARE.

 </details>

-### apprise (1.9.7) — BSD-2-Clause
+### apprise (1.9.9) — BSD-2-Clause

 <details>
 <summary>Full license text</summary>
@@ -64,7 +64,7 @@ SOFTWARE.
 ```
 BSD 2-Clause License

-Copyright (c) 2025, Chris Caron <lead2gold@gmail.com>
+Copyright (c) 2026, Chris Caron <lead2gold@gmail.com>
 All rights reserved.

 Redistribution and use in source and binary forms, with or without
@@ -1188,6 +1188,37 @@ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLI

 </details>

+### cmdk (1.1.1) — MIT
+
+<details>
+<summary>Full license text</summary>
+
+```
+MIT License
+
+Copyright (c) 2022 Paco Coursey
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+```
+
+</details>
+
 ### d3-force (3.0.0) — ISC

 <details>
@@ -12,25 +12,18 @@ Backend server + browser interface for MeshCore mesh radio networks. Connect you
 * Use the more recent 1.14 firmwares which support multibyte pathing
 * Visualize the mesh as a map or node set, view repeater stats, and more!

+For advanced setup and troubleshooting see [README_ADVANCED.md](README_ADVANCED.md). If you plan to contribute, read [CONTRIBUTING.md](CONTRIBUTING.md).
+
 **Warning:** This app is for trusted environments only. _Do not put this on an untrusted network, or open it to the public._ You can optionally set `MESHCORE_BASIC_AUTH_USERNAME` and `MESHCORE_BASIC_AUTH_PASSWORD` for app-wide HTTP Basic auth, but that is only a coarse gate and must be paired with HTTPS. The bots can execute arbitrary Python code which means anyone who gets access to the app can, too. To completely disable the bot system, start the server with `MESHCORE_DISABLE_BOTS=true` — this prevents all bot execution and blocks bot configuration changes via the API. If you need stronger access control, consider using a reverse proxy like Nginx, or extending FastAPI; full access control and user management are outside the scope of this app.

 ![Screenshot of the application's web interface](app_screenshot.png)

-## Start Here
-
-Most users should choose one of these paths:
-
-1. Clone and build from source.
-2. Download the prebuilt release zip if you are on a resource-constrained system and do not want to build the frontend locally.
-3. Use Docker if that better matches how you deploy.
-
-For advanced setup, troubleshooting, HTTPS, systemd service setup, and remediation environment variables, see [README_ADVANCED.md](README_ADVANCED.md).
-
-If you plan to contribute, read [CONTRIBUTING.md](CONTRIBUTING.md).
+> [!WARNING]
+> RemoteTerm does *full* management of the radio, meaning that once a radio is connected to RemoteTerm, all contacts/channels will be imported and offloaded to RemoteTerm and the contacts actually synced to the device will be governed by RemoteTerm. This means that RemoteTerm can be a poor fit for users who are looking to swap radios in and out, maintaining radio state (favorites, channels, etc.) irrespective of app usage.

 ## Requirements

- Python 3.10+
+- Python 3.11+
 - Node.js LTS or current (20, 22, 24, 25) if you're not using a prebuilt release
 - [UV](https://astral.sh/uv) package manager: `curl -LsSf https://astral.sh/uv/install.sh | sh`
 - MeshCore radio connected via USB serial, TCP, or BLE
@@ -71,7 +64,7 @@ usbipd attach --wsl --busid 3-8
 ```
 </details>

-## Path 1: Clone And Build
+## Install Path 1: Clone And Build

 **This approach is recommended over Docker due to intermittent serial communications issues I've seen on \*nix systems.**

@@ -89,10 +82,10 @@ Access the app at http://localhost:8000.

 Source checkouts expect a normal frontend build in `frontend/dist`.

-> [!NOTE]
+> [!TIP]
 > Running on lightweight hardware, or just do not want to build the frontend locally? From a cloned checkout, run `python3 scripts/setup/fetch_prebuilt_frontend.py` to fetch and unpack a prebuilt frontend into `frontend/prebuilt`, then start the app normally with `uv run uvicorn app.main:app --host 0.0.0.0 --port 8000`.

-> [!TIP]
+> [!NOTE]
 > On Linux, you can also install RemoteTerm as a persistent `systemd` service that starts on boot and restarts automatically on failure:
 >
 > ```bash
@@ -101,7 +94,7 @@ Source checkouts expect a normal frontend build in `frontend/dist`.
 >
 > For the full service workflow and post-install operations, see [README_ADVANCED.md](README_ADVANCED.md).

-## Path 2: Docker
+## Install Path 2: Docker

 > **Warning:** Docker has had reports intermittent issues with serial event subscriptions. The native method above is more reliable.

@@ -123,6 +116,8 @@ cp docker-compose.example.yml docker-compose.yml
 bash scripts/setup/install_docker.sh
 ```

+> The interactive generator enables a self-signed (snakeoil) TLS certificate by default. If you accept the default, the app will be served over HTTPS and the generated compose file will include certificate mounts and an SSL command override. Decline if you prefer plain HTTP or plan to terminate TLS externally.
+
 Your local `docker-compose.yml` is gitignored so future pulls do not overwrite your Docker settings.

 The guided Docker flow can collect BLE settings, but BLE access from Docker still needs manual compose customization such as Bluetooth passthrough and possibly privileged mode or host networking. If you want the simpler path for BLE, use the regular Python launch flow instead.
@@ -142,6 +137,8 @@ sudo docker compose pull
 sudo docker compose up -d
 ```

+> If you switched to a local build (`build: .` instead of `image:`), use `sudo docker compose up -d --build` instead — `pull` only fetches remote images.
+
 The example file and setup script default to the published Docker Hub image. To build locally from your checkout instead, replace:

 ```yaml
@@ -168,6 +165,29 @@ To stop:
 sudo docker compose down
 ```

+## Install Path 3: Arch Linux (AUR)
+
+A [`remoteterm-meshcore`](https://aur.archlinux.org/packages/remoteterm-meshcore) package is available in the AUR. Install it with an AUR helper or build it manually:
+
+```bash
+# with an AUR helper
+yay -S remoteterm-meshcore
+
+# or manually
+git clone https://aur.archlinux.org/remoteterm-meshcore.git
+cd remoteterm-meshcore
+makepkg -si
+```
+
+Configure your radio connection, then start the service:
+
+```bash
+sudo vi /etc/remoteterm-meshcore/remoteterm.env
+sudo systemctl enable --now remoteterm-meshcore
+```
+
+Access the app at http://localhost:8000.
+
 ## Standard Environment Variables

 Only one transport may be active at a time. If multiple are set, the server will refuse to start.
@@ -177,7 +197,7 @@ Only one transport may be active at a time. If multiple are set, the server will
 | `MESHCORE_SERIAL_PORT` | (auto-detect) | Serial port path |
 | `MESHCORE_SERIAL_BAUDRATE` | 115200 | Serial baud rate |
 | `MESHCORE_TCP_HOST` | | TCP host (mutually exclusive with serial/BLE) |
-| `MESHCORE_TCP_PORT` | 4000 | TCP port |
+| `MESHCORE_TCP_PORT` | 5000 | TCP port |
 | `MESHCORE_BLE_ADDRESS` | | BLE device address (mutually exclusive with serial/TCP) |
 | `MESHCORE_BLE_PIN` | | BLE PIN (required when BLE address is set) |
 | `MESHCORE_LOG_LEVEL` | INFO | `DEBUG`, `INFO`, `WARNING`, `ERROR` |
@@ -193,7 +213,7 @@ Common launch patterns:
 MESHCORE_SERIAL_PORT=/dev/ttyUSB0 uv run uvicorn app.main:app --host 0.0.0.0 --port 8000

 # TCP
-MESHCORE_TCP_HOST=192.168.1.100 MESHCORE_TCP_PORT=4000 uv run uvicorn app.main:app --host 0.0.0.0 --port 8000
+MESHCORE_TCP_HOST=192.168.1.100 MESHCORE_TCP_PORT=5000 uv run uvicorn app.main:app --host 0.0.0.0 --port 8000

 # BLE
 MESHCORE_BLE_ADDRESS=AA:BB:CC:DD:EE:FF MESHCORE_BLE_PIN=123456 uv run uvicorn app.main:app --host 0.0.0.0 --port 8000
@@ -206,6 +226,15 @@ $env:MESHCORE_SERIAL_PORT="COM8" # or your COM port
 uv run uvicorn app.main:app --host 0.0.0.0 --port 8000
 ```

+> [!WARNING]
+> **Windows + MQTT fanout:** Python's default Windows event loop (ProactorEventLoop) is not compatible with the MQTT libraries used by RemoteTerm. If you configure any MQTT integration, add `--loop none` to your uvicorn command:
+>
+> ```powershell
+> uv run uvicorn app.main:app --host 0.0.0.0 --port 8000 --loop none
+> ```
+>
+> If you forget, the app will start normally but MQTT connections will fail and you'll see a toast in the UI with this same guidance.
+
 If you enable Basic Auth, protect the app with HTTPS. HTTP Basic credentials are not safe on plain HTTP. Also note that the app's permissive CORS policy is a deliberate trusted-network tradeoff, so cross-origin browser JavaScript is not a reliable way to use that Basic Auth gate.

 ## Where To Go Next
@@ -19,6 +19,15 @@ If the audit finds a mismatch, you'll see an error in the application UI and you

 `__CLOWNTOWN_DO_CLOCK_WRAPAROUND=true` is a last-resort clock remediation for nodes whose RTC is stuck in the future and where rescue-mode time setting or GPS-based time is not available. It intentionally relies on the clock rolling past the 32-bit epoch boundary, which is board-specific behavior and may not be safe or effective on all MeshCore targets. Treat it as highly experimental.

+## Sub-Path Reverse Proxy
+
+RemoteTerm works behind a reverse proxy that serves it under a sub-path (e.g. `/meshcore/` or Home Assistant ingress). All frontend asset and API paths are relative, so they resolve correctly under any prefix.
+
+**Requirements:**
+
+- The proxy must ensure the sub-path URL has a **trailing slash**. If a user visits `/meshcore` (no slash), relative paths break. Most proxies handle this automatically; for Nginx, a `location /meshcore/ { ... }` block (note the trailing slash) does the right thing.
+- For correct PWA install behavior, the proxy should forward `X-Forwarded-Prefix` (set to the sub-path, e.g. `/meshcore`) so the web manifest generates correct `start_url` and `scope` values. `X-Forwarded-Proto` and `X-Forwarded-Host` are also respected for origin resolution.
+
 ## HTTPS

 WebGPU channel-finding requires a secure context when you are not on `localhost`.
@@ -0,0 +1,305 @@
+# Home Assistant Integration
+
+RemoteTerm can publish mesh network data to Home Assistant via MQTT Discovery. Devices and entities appear automatically in HA -- no custom component or HACS install needed.
+
+## Prerequisites
+
+- Home Assistant with the [MQTT integration](https://www.home-assistant.io/integrations/mqtt/) configured
+- An MQTT broker (e.g. Mosquitto) accessible to both HA and RemoteTerm
+- RemoteTerm running and connected to a radio
+
+## Setup
+
+1. In RemoteTerm, go to **Settings > Integrations > Add > Home Assistant MQTT Discovery**
+2. Enter your MQTT broker host and port (same broker HA is connected to)
+3. Optionally enter broker username/password and TLS settings
+4. Select contacts for GPS tracking and repeaters for telemetry (see below)
+5. Configure which messages should fire events (scope selector at the bottom)
+6. Save and enable
+
+Devices will appear in HA under **Settings > Devices & Services > MQTT** within a few seconds.
+
+## What Gets Created
+
+### Local Radio Device
+
+Always created. Updates every 60 seconds.
+
+| Entity | Type | Description |
+|--------|------|-------------|
+| `binary_sensor.meshcore_*_connected` | Connectivity | Radio online/offline |
+| `sensor.meshcore_*_noise_floor` | Signal strength | Radio noise floor (dBm) |
+
+### Repeater Devices
+
+One device per tracked repeater (must have repeater opted). Updates when telemetry is collected (auto-collect cycle (~8 hours), or when you manually fetch from the repeater dashboard).
+
+Repeaters must first be added to the auto-telemetry tracking list in RemoteTerm's Radio settings section. Only auto-tracked repeaters appear in the HA integration's repeater picker.
+
+| Entity | Type | Unit | Description |
+|--------|------|------|-------------|
+| `sensor.meshcore_*_battery_voltage` | Voltage | V | Battery level |
+| `sensor.meshcore_*_noise_floor` | Signal strength | dBm | Local noise floor |
+| `sensor.meshcore_*_last_rssi` | Signal strength | dBm | Last received signal strength |
+| `sensor.meshcore_*_last_snr` | -- | dB | Last signal-to-noise ratio |
+| `sensor.meshcore_*_packets_received` | -- | count | Total packets received |
+| `sensor.meshcore_*_packets_sent` | -- | count | Total packets sent |
+| `sensor.meshcore_*_uptime` | Duration | s | Uptime since last reboot |
+
+### Contact Device Trackers
+
+One `device_tracker` per tracked contact. Updates passively whenever RemoteTerm hears an advertisement with GPS coordinates from that contact. No radio commands are sent -- it piggybacks on normal mesh traffic.
+
+| Entity | Description |
+|--------|-------------|
+| `device_tracker.meshcore_*` | GPS position (latitude/longitude) |
+
+### Message Event Entity
+
+A single `event.meshcore_messages` entity that fires for each message matching your configured scope. Each event carries these attributes:
+
+| Attribute | Example | Description |
+|-----------|---------|-------------|
+| `event_type` | `message_received` | Always `message_received` |
+| `sender_name` | `Alice` | Display name of the sender |
+| `sender_key` | `aabbccdd...` | Sender's public key |
+| `text` | `hello` | Message body |
+| `message_type` | `PRIV` or `CHAN` | Direct message or channel |
+| `channel_name` | `#general` | Channel name (null for DMs) |
+| `conversation_key` | `aabbccdd...` | Contact key (DM) or channel key |
+| `outgoing` | `false` | Whether you sent this message |
+
+## Entity Naming
+
+Entity IDs use the first 12 characters of the node's public key as an identifier. For example, a contact with public key `ae92577bae6c...` gets entity ID `device_tracker.meshcore_ae92577bae6c`. You can rename entities in HA's UI without affecting the integration.
+
+## Common Automations
+
+### Low repeater battery alert
+
+Notify when a tracked repeater's battery drops below a threshold.
+
+**GUI:** Settings > Automations > Create > Numeric state trigger on `sensor.meshcore_*_battery_voltage`, below `3.8`, action: notification.
+
+**YAML:**
+```yaml
+automation:
+  - alias: "Repeater battery low"
+    trigger:
+      - platform: numeric_state
+        entity_id: sensor.meshcore_aabbccddeeff_battery_voltage
+        below: 3.8
+    action:
+      - service: notify.mobile_app_your_phone
+        data:
+          title: "Repeater Battery Low"
+          message: >-
+            {{ state_attr('sensor.meshcore_aabbccddeeff_battery_voltage', 'friendly_name') }}
+            is at {{ states('sensor.meshcore_aabbccddeeff_battery_voltage') }}V
+```
+
+### Radio offline alert
+
+Notify if the radio has been disconnected for more than 5 minutes.
+
+**GUI:** Settings > Automations > Create > State trigger on `binary_sensor.meshcore_*_connected`, to `off`, for `00:05:00`, action: notification.
+
+**YAML:**
+```yaml
+automation:
+  - alias: "Radio offline"
+    trigger:
+      - platform: state
+        entity_id: binary_sensor.meshcore_aabbccddeeff_connected
+        to: "off"
+        for: "00:05:00"
+    action:
+      - service: notify.mobile_app_your_phone
+        data:
+          title: "MeshCore Radio Offline"
+          message: "Radio has been disconnected for 5 minutes"
+```
+
+### Alert on any message from a specific room
+
+Trigger when a message arrives in a specific channel. Two approaches:
+
+#### Option A: Scope filtering (fully GUI, no template)
+
+If you only care about one room, configure the HA integration's message scope to "Only listed channels" and select that room. Then every event fire is from that room.
+
+**GUI:** Settings > Automations > Create > State trigger on `event.meshcore_messages`, action: notification.
+
+**YAML:**
+```yaml
+automation:
+  - alias: "Emergency channel alert"
+    trigger:
+      - platform: state
+        entity_id: event.meshcore_messages
+    action:
+      - service: notify.mobile_app_your_phone
+        data:
+          title: "Message in #emergency"
+          message: >-
+            {{ trigger.to_state.attributes.sender_name }}:
+            {{ trigger.to_state.attributes.text }}
+```
+
+#### Option B: Template condition (multiple rooms, one integration)
+
+Keep scope as "All messages" and filter in the automation. The trigger is GUI, but the condition uses a one-line template.
+
+**GUI:** Settings > Automations > Create > State trigger on `event.meshcore_messages` > Add condition > Template > enter the template below.
+
+**YAML:**
+```yaml
+automation:
+  - alias: "Emergency channel alert"
+    trigger:
+      - platform: state
+        entity_id: event.meshcore_messages
+    condition:
+      - condition: template
+        value_template: >-
+          {{ trigger.to_state.attributes.channel_name == '#emergency' }}
+    action:
+      - service: notify.mobile_app_your_phone
+        data:
+          title: "Message in #emergency"
+          message: >-
+            {{ trigger.to_state.attributes.sender_name }}:
+            {{ trigger.to_state.attributes.text }}
+```
+
+### Alert on DM from a specific contact
+
+**YAML:**
+```yaml
+automation:
+  - alias: "DM from Alice"
+    trigger:
+      - platform: state
+        entity_id: event.meshcore_messages
+    condition:
+      - condition: template
+        value_template: >-
+          {{ trigger.to_state.attributes.message_type == 'PRIV'
+             and trigger.to_state.attributes.sender_name == 'Alice' }}
+    action:
+      - service: notify.mobile_app_your_phone
+        data:
+          title: "DM from Alice"
+          message: "{{ trigger.to_state.attributes.text }}"
+```
+
+### Alert on messages containing a keyword
+
+**YAML:**
+```yaml
+automation:
+  - alias: "Keyword alert"
+    trigger:
+      - platform: state
+        entity_id: event.meshcore_messages
+    condition:
+      - condition: template
+        value_template: >-
+          {{ 'emergency' in trigger.to_state.attributes.text | lower }}
+    action:
+      - service: notify.mobile_app_your_phone
+        data:
+          title: "Emergency keyword detected"
+          message: >-
+            {{ trigger.to_state.attributes.sender_name }} in
+            {{ trigger.to_state.attributes.channel_name or 'DM' }}:
+            {{ trigger.to_state.attributes.text }}
+```
+
+### Track a contact on the HA map
+
+No automation needed. Once a contact is selected for GPS tracking, their `device_tracker` entity automatically appears on the HA map. Go to **Settings > Dashboards > Map** (or add a Map card to any dashboard) and the tracked contact will show up when they advertise their GPS position.
+
+### Dashboard card showing repeater battery
+
+Add a sensor card to any dashboard:
+
+```yaml
+type: sensor
+entity: sensor.meshcore_aabbccddeeff_battery_voltage
+name: "Hilltop Repeater Battery"
+```
+
+Or an entities card for multiple repeaters:
+
+```yaml
+type: entities
+title: "Repeater Status"
+entities:
+  - entity: sensor.meshcore_aabbccddeeff_battery_voltage
+    name: "Hilltop"
+  - entity: sensor.meshcore_ccdd11223344_battery_voltage
+    name: "Valley"
+  - entity: sensor.meshcore_eeff55667788_battery_voltage
+    name: "Ridge"
+```
+
+## Troubleshooting
+
+### Devices don't appear in HA
+
+- Verify the MQTT integration is configured in HA (**Settings > Devices & Services > MQTT**) and shows "Connected"
+- Verify RemoteTerm's HA integration shows "Connected" (green dot)
+- Check that both HA and RemoteTerm are using the same MQTT broker
+- Subscribe to discovery topics to verify messages are flowing:
+  ```
+  mosquitto_sub -h <broker> -t 'homeassistant/#' -v
+  ```
+
+### Stale or duplicate devices
+
+If you see unexpected devices (e.g. a generic "MeshCore Radio" alongside your named radio), clear the stale retained messages:
+```
+mosquitto_pub -h <broker> -t 'homeassistant/binary_sensor/meshcore_unknown/connected/config' -r -n
+mosquitto_pub -h <broker> -t 'homeassistant/sensor/meshcore_unknown/noise_floor/config' -r -n
+```
+
+### Repeater sensors show "Unknown" or "Unavailable"
+
+Repeater telemetry only updates when collected. Trigger a manual fetch by opening the repeater's dashboard in RemoteTerm and clicking "Status", or wait for the next auto-collect cycle (~8 hours). Sensors show "Unknown" until the first telemetry reading arrives.
+
+### Contact device tracker shows "Unknown"
+
+The contact's GPS position only updates when RemoteTerm hears an advertisement from that node that includes GPS coordinates. If the contact's device doesn't broadcast GPS or hasn't advertised recently, the tracker will show as unknown.
+
+### Entity is "Unavailable"
+
+Radio health entities have a 120-second expiry. If RemoteTerm stops sending health updates (e.g. it's shut down or loses connection to the broker), HA marks the entities as unavailable after 2 minutes. Restart RemoteTerm or check the broker connection.
+
+## Removing the Integration
+
+Disabling or deleting the HA integration in RemoteTerm's settings publishes empty retained messages to all discovery topics, which removes the devices and entities from HA automatically.
+
+## MQTT Topics Reference
+
+State topics (where data is published):
+
+| Topic | Content | Update frequency |
+|-------|---------|-----------------|
+| `meshcore/{node_id}/health` | `{"connected": bool, "noise_floor_dbm": int}` | Every 60s |
+| `meshcore/{node_id}/telemetry` | `{"battery_volts": float, ...}` | ~8h or manual |
+| `meshcore/{node_id}/gps` | `{"latitude": float, "longitude": float, ...}` | On advert |
+| `meshcore/events/message` | `{"event_type": "message_received", ...}` | On message |
+
+Discovery topics (entity registration, under `homeassistant/`):
+
+| Pattern | Entity type |
+|---------|------------|
+| `homeassistant/binary_sensor/meshcore_*/connected/config` | Radio connectivity |
+| `homeassistant/sensor/meshcore_*/noise_floor/config` | Noise floor sensor |
+| `homeassistant/sensor/meshcore_*/battery_voltage/config` | Repeater battery |
+| `homeassistant/sensor/meshcore_*/*/config` | Other repeater sensors |
+| `homeassistant/device_tracker/meshcore_*/config` | Contact GPS tracker |
+| `homeassistant/event/meshcore_messages/config` | Message event entity |
+
+The `{node_id}` is always the first 12 characters of the node's public key, lowercased.
@@ -40,7 +40,7 @@ app/
 │   ├── contact_reconciliation.py # Prefix-claim, sender-key backfill, name-history wiring
 │   ├── radio_lifecycle.py       # Post-connect setup and reconnect/setup helpers
 │   ├── radio_commands.py        # Radio config/private-key command workflows
-│   ├── radio_noise_floor.py     # In-memory local radio noise-floor sampling/history
+│   ├── radio_stats.py           # In-memory local radio stats sampling and noise-floor history
 │   └── radio_runtime.py         # Router/dependency seam over the global RadioManager
 ├── radio.py             # RadioManager transport/session state + lock management
 ├── radio_sync.py        # Polling, sync, periodic advertisement loop
@@ -161,10 +161,12 @@ app/

 - All external integrations (MQTT, bots, webhooks, Apprise, SQS) are managed through the fanout bus (`app/fanout/`).
 - Configs stored in `fanout_configs` table, managed via `GET/POST/PATCH/DELETE /api/fanout`.
- `broadcast_event()` in `websocket.py` dispatches to the fanout manager for `message` and `raw_packet` events.
- Each integration is a `FanoutModule` with scope-based filtering.
+- `broadcast_event()` in `websocket.py` dispatches to the fanout manager for `message`, `raw_packet`, and `contact` events.
+- `on_message` and `on_raw` are scope-gated. `on_contact`, `on_telemetry`, and `on_health` are dispatched to all modules unconditionally (modules filter internally).
+- Repeater telemetry broadcasts are emitted after `RepeaterTelemetryRepository.record()` in both `radio_sync.py` (auto-collect) and `routers/repeaters.py` (manual fetch).
+- The 60-second radio stats sampling loop in `radio_stats.py` dispatches an enriched health snapshot (radio identity + full stats) to all fanout modules after each sample.
 - Community MQTT publishes raw packets only, but its derived `path` field for direct packets is emitted as comma-separated hop identifiers, not flat path bytes.
- See `app/fanout/AGENTS_fanout.md` for full architecture details.
+- See `app/fanout/AGENTS_fanout.md` for full architecture details and event payload shapes.

 ## API Surface (all under `/api`)

@@ -190,6 +192,7 @@ app/
 - `GET /contacts/analytics` — unified keyed-or-name analytics payload
 - `GET /contacts/repeaters/advert-paths` — recent advert paths for all contacts
 - `POST /contacts`
+- `POST /contacts/bulk-delete`
 - `DELETE /contacts/{public_key}`
 - `POST /contacts/{public_key}/mark-read`
 - `POST /contacts/{public_key}/command`
@@ -214,8 +217,10 @@ app/
 - `GET /channels`
 - `GET /channels/{key}/detail`
 - `POST /channels`
+- `POST /channels/bulk-hashtag`
 - `DELETE /channels/{key}`
 - `POST /channels/{key}/flood-scope-override`
+- `POST /channels/{key}/path-hash-mode-override`
 - `POST /channels/{key}/mark-read`

 ### Messages
@@ -241,7 +246,7 @@ app/
 - `POST /settings/favorites/toggle`
 - `POST /settings/blocked-keys/toggle`
 - `POST /settings/blocked-names/toggle`
- `POST /settings/migrate`
+- `POST /settings/tracked-telemetry/toggle`

 ### Fanout
 - `GET /fanout` — list all fanout configs
@@ -278,11 +283,13 @@ Client sends `"ping"` text; server replies `{"type":"pong"}`.
 Main tables:
 - `contacts` (includes `first_seen` for contact age tracking and `direct_path_hash_mode` / `route_override_*` for DM routing)
 - `channels`
-  Includes optional `flood_scope_override` for channel-specific regional sends.
+  Includes optional `flood_scope_override` for channel-specific regional sends and optional `path_hash_mode_override` for per-channel path hop width.
 - `messages` (includes `sender_name`, `sender_key` for per-contact channel message attribution)
 - `raw_packets`
 - `contact_advert_paths` (recent unique advertisement paths per contact, keyed by contact + path bytes + hop count)
 - `contact_name_history` (tracks name changes over time)
+- `repeater_telemetry_history` (time-series telemetry snapshots for tracked repeaters)
+- `fanout_configs` (MQTT, bot, webhook, Apprise, SQS integration configs)
 - `app_settings`

 Contact route state is canonicalized on the backend:
@@ -298,17 +305,14 @@ Repository writes should prefer typed models such as `ContactUpsert` over ad hoc

 `app_settings` fields in active model:
 - `max_radio_contacts`
- `favorites`
 - `auto_decrypt_dm_on_advert`
- `sidebar_sort_order`
 - `last_message_times`
- `preferences_migrated`
 - `advert_interval`
 - `last_advert_time`
 - `flood_scope`
- `blocked_keys`, `blocked_names`
-
-Note: `sidebar_sort_order` remains in the backend model for compatibility and migration, but the current frontend sidebar uses per-section localStorage sort preferences instead of a single shared server-backed sort mode.
+- `blocked_keys`, `blocked_names`, `discovery_blocked_types`
+- `tracked_telemetry_repeaters`
+- `auto_resend_channel`

 Note: MQTT, community MQTT, and bot configs were migrated to the `fanout_configs` table (migrations 36-38).

@@ -14,7 +14,7 @@ class Settings(BaseSettings):
    serial_port: str = ""  # Empty string triggers auto-detection
    serial_baudrate: int = 115200
    tcp_host: str = ""
-    tcp_port: int = 4000
+    tcp_port: int = 5000
    ble_address: str = ""
    ble_pin: str = ""
    log_level: Literal["DEBUG", "INFO", "WARNING", "ERROR"] = "INFO"
@@ -26,6 +26,7 @@ class Settings(BaseSettings):
        default=False,
        validation_alias="__CLOWNTOWN_DO_CLOCK_WRAPAROUND",
    )
+    skip_post_connect_sync: bool = False
    basic_auth_username: str = ""
    basic_auth_password: str = ""

@@ -7,7 +7,7 @@ from app.config import settings

 logger = logging.getLogger(__name__)

-SCHEMA = """
+SCHEMA_TABLES = """
 CREATE TABLE IF NOT EXISTS contacts (
    public_key TEXT PRIMARY KEY,
    name TEXT,
@@ -27,7 +27,8 @@ CREATE TABLE IF NOT EXISTS contacts (
    on_radio INTEGER DEFAULT 0,
    last_contacted INTEGER,
    first_seen INTEGER,
-    last_read_at INTEGER
+    last_read_at INTEGER,
+    favorite INTEGER DEFAULT 0
 );

 CREATE TABLE IF NOT EXISTS channels (
@@ -36,7 +37,9 @@ CREATE TABLE IF NOT EXISTS channels (
    is_hashtag INTEGER DEFAULT 0,
    on_radio INTEGER DEFAULT 0,
    flood_scope_override TEXT,
-    last_read_at INTEGER
+    path_hash_mode_override INTEGER,
+    last_read_at INTEGER,
+    favorite INTEGER DEFAULT 0
 );

 CREATE TABLE IF NOT EXISTS messages (
@@ -46,7 +49,7 @@ CREATE TABLE IF NOT EXISTS messages (
    text TEXT NOT NULL,
    sender_timestamp INTEGER,
    received_at INTEGER NOT NULL,
-    path TEXT,
+    paths TEXT,
    txt_type INTEGER DEFAULT 0,
    signature TEXT,
    outgoing INTEGER DEFAULT 0,
@@ -91,23 +94,72 @@ CREATE TABLE IF NOT EXISTS contact_name_history (
    FOREIGN KEY (public_key) REFERENCES contacts(public_key) ON DELETE CASCADE
 );

+CREATE TABLE IF NOT EXISTS app_settings (
+    id INTEGER PRIMARY KEY CHECK (id = 1),
+    max_radio_contacts INTEGER DEFAULT 200,
+    favorites TEXT DEFAULT '[]',
+    auto_decrypt_dm_on_advert INTEGER DEFAULT 1,
+    last_message_times TEXT DEFAULT '{}',
+    preferences_migrated INTEGER DEFAULT 0,
+    advert_interval INTEGER DEFAULT 0,
+    last_advert_time INTEGER DEFAULT 0,
+    flood_scope TEXT DEFAULT '',
+    blocked_keys TEXT DEFAULT '[]',
+    blocked_names TEXT DEFAULT '[]',
+    discovery_blocked_types TEXT DEFAULT '[]',
+    tracked_telemetry_repeaters TEXT DEFAULT '[]',
+    auto_resend_channel INTEGER DEFAULT 0
+);
+INSERT OR IGNORE INTO app_settings (id) VALUES (1);
+
+CREATE TABLE IF NOT EXISTS fanout_configs (
+    id TEXT PRIMARY KEY,
+    type TEXT NOT NULL,
+    name TEXT NOT NULL,
+    enabled INTEGER DEFAULT 0,
+    config TEXT NOT NULL DEFAULT '{}',
+    scope TEXT NOT NULL DEFAULT '{}',
+    sort_order INTEGER DEFAULT 0,
+    created_at INTEGER NOT NULL
+);
+
+CREATE TABLE IF NOT EXISTS repeater_telemetry_history (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    public_key TEXT NOT NULL,
+    timestamp INTEGER NOT NULL,
+    data TEXT NOT NULL,
+    FOREIGN KEY (public_key) REFERENCES contacts(public_key) ON DELETE CASCADE
+);
+"""
+
+# Indexes are created after migrations so that legacy databases have all
+# required columns (e.g. sender_key, added by migration 25) before index
+# creation runs.
+SCHEMA_INDEXES = """
 CREATE INDEX IF NOT EXISTS idx_messages_received ON messages(received_at);
 CREATE UNIQUE INDEX IF NOT EXISTS idx_messages_dedup_null_safe
    ON messages(type, conversation_key, text, COALESCE(sender_timestamp, 0))
    WHERE type = 'CHAN';
+CREATE UNIQUE INDEX IF NOT EXISTS idx_messages_incoming_priv_dedup
+    ON messages(type, conversation_key, text, COALESCE(sender_timestamp, 0), COALESCE(sender_key, ''))
+    WHERE type = 'PRIV' AND outgoing = 0;
+CREATE INDEX IF NOT EXISTS idx_messages_sender_key ON messages(sender_key);
+CREATE INDEX IF NOT EXISTS idx_messages_pagination
+    ON messages(type, conversation_key, received_at DESC, id DESC);
+CREATE INDEX IF NOT EXISTS idx_messages_unread_covering
+    ON messages(type, conversation_key, outgoing, received_at);
 CREATE INDEX IF NOT EXISTS idx_raw_packets_message_id ON raw_packets(message_id);
 CREATE INDEX IF NOT EXISTS idx_raw_packets_timestamp ON raw_packets(timestamp);
 CREATE UNIQUE INDEX IF NOT EXISTS idx_raw_packets_payload_hash ON raw_packets(payload_hash);
-CREATE INDEX IF NOT EXISTS idx_contacts_on_radio ON contacts(on_radio);
 CREATE INDEX IF NOT EXISTS idx_contacts_type_last_seen ON contacts(type, last_seen);
 CREATE INDEX IF NOT EXISTS idx_messages_type_received_conversation
    ON messages(type, received_at, conversation_key);
-- idx_messages_sender_key is created by migration 25 (after adding the sender_key column)
-- idx_messages_incoming_priv_dedup is created by migration 44 after legacy rows are reconciled
 CREATE INDEX IF NOT EXISTS idx_contact_advert_paths_recent
    ON contact_advert_paths(public_key, last_seen DESC);
 CREATE INDEX IF NOT EXISTS idx_contact_name_history_key
    ON contact_name_history(public_key, last_seen DESC);
+CREATE INDEX IF NOT EXISTS idx_repeater_telemetry_pk_ts
+    ON repeater_telemetry_history(public_key, timestamp);
 """


@@ -126,6 +178,22 @@ class Database:
        # Persists in the DB file but we set it explicitly on every connection.
        await self._connection.execute("PRAGMA journal_mode = WAL")

+        # synchronous = NORMAL is safe with WAL — only the most recent
+        # transaction can be lost on an OS crash (no corruption risk).
+        # Reduces fsync overhead vs. the default FULL.
+        await self._connection.execute("PRAGMA synchronous = NORMAL")
+
+        # Retry for up to 5s on lock contention instead of failing instantly.
+        # Matters when a second connection (e.g. VACUUM) touches the DB.
+        await self._connection.execute("PRAGMA busy_timeout = 5000")
+
+        # Bump page cache to ~64 MB (negative value = KB). Keeps hot pages
+        # in memory for read-heavy queries (unreads, pagination, search).
+        await self._connection.execute("PRAGMA cache_size = -64000")
+
+        # Keep temp tables and sort spills in memory instead of on disk.
+        await self._connection.execute("PRAGMA temp_store = MEMORY")
+
        # Incremental auto-vacuum: freed pages are reclaimable via
        # PRAGMA incremental_vacuum without a full VACUUM. Must be set before
        # the first table is created (for new databases); for existing databases
@@ -138,15 +206,20 @@ class Database:
        # constraints, then re-enabled for all subsequent application queries.
        await self._connection.execute("PRAGMA foreign_keys = OFF")

-        await self._connection.executescript(SCHEMA)
+        await self._connection.executescript(SCHEMA_TABLES)
        await self._connection.commit()
-        logger.debug("Database schema initialized")
+        logger.debug("Database tables initialized")

-        # Run any pending migrations
+        # Run any pending migrations before creating indexes, so that
+        # legacy databases have all required columns first.
        from app.migrations import run_migrations

        await run_migrations(self._connection)

+        await self._connection.executescript(SCHEMA_INDEXES)
+        await self._connection.commit()
+        logger.debug("Database indexes initialized")
+
        # Enable FK enforcement for all application queries from this point on.
        await self._connection.execute("PRAGMA foreign_keys = ON")
        logger.debug("Foreign key enforcement enabled")
@@ -299,8 +299,11 @@ def parse_advertisement(
    timestamp = int.from_bytes(payload[32:36], byteorder="little")
    flags = payload[100]

-    # Parse flags
+    # Parse flags — clamp device_role to valid range (0-4); corrupted
+    # advertisements can have junk in the lower nibble.
    device_role = flags & 0x0F
+    if device_role > 4:
+        device_role = 0
    has_location = bool(flags & 0x10)
    has_feature1 = bool(flags & 0x20)
    has_feature2 = bool(flags & 0x40)
@@ -1,8 +0,0 @@
-"""Shared dependencies for FastAPI routers."""
-
-from app.services.radio_runtime import radio_runtime as radio_manager
-
-
-def require_connected():
-    """Dependency that ensures radio is connected and returns meshcore instance."""
-    return radio_manager.require_connected()
@@ -202,7 +202,6 @@ async def on_path_update(event: "Event") -> None:
        # Legacy firmware/library payloads only support 1-byte hop hashes.
        normalized_path_hash_mode = -1 if normalized_path_len == -1 else 0
    else:
-        normalized_path_hash_mode = None
        try:
            normalized_path_hash_mode = int(path_hash_mode)
        except (TypeError, ValueError):
@@ -2,10 +2,10 @@

 import json
 import logging
-from typing import Any, Literal
+from typing import Any, Literal, NotRequired

 from pydantic import TypeAdapter
-from typing_extensions import NotRequired, TypedDict
+from typing_extensions import TypedDict

 from app.models import Channel, Contact, Message, MessagePath, RawPacketBroadcast
 from app.routers.health import HealthResponse
@@ -52,19 +52,6 @@ class ToastPayload(TypedDict):
    details: NotRequired[str]


-WsEventPayload = (
-    HealthResponse
-    | Message
-    | Contact
-    | ContactResolvedPayload
-    | Channel
-    | ContactDeletedPayload
-    | ChannelDeletedPayload
-    | RawPacketBroadcast
-    | MessageAckedPayload
-    | ToastPayload
-)
-
 _PAYLOAD_ADAPTERS: dict[WsEventType, TypeAdapter[Any]] = {
    "health": TypeAdapter(HealthResponse),
    "message": TypeAdapter(Message),
@@ -80,14 +67,6 @@ _PAYLOAD_ADAPTERS: dict[WsEventType, TypeAdapter[Any]] = {
 }


-def validate_ws_event_payload(event_type: str, data: Any) -> WsEventPayload | Any:
-    """Validate known WebSocket payloads; pass unknown events through unchanged."""
-    adapter = _PAYLOAD_ADAPTERS.get(event_type)  # type: ignore[arg-type]
-    if adapter is None:
-        return data
-    return adapter.validate_python(data)
-
-
 def dump_ws_event(event_type: str, data: Any) -> str:
    """Serialize a WebSocket event envelope with validation for known event types."""
    adapter = _PAYLOAD_ADAPTERS.get(event_type)  # type: ignore[arg-type]
@@ -104,13 +83,3 @@ def dump_ws_event(event_type: str, data: Any) -> str:
            event_type,
        )
        return json.dumps({"type": event_type, "data": data})
-
-
-def dump_ws_event_payload(event_type: str, data: Any) -> Any:
-    """Return the JSON-serializable payload for a WebSocket event."""
-    adapter = _PAYLOAD_ADAPTERS.get(event_type)  # type: ignore[arg-type]
-    if adapter is None:
-        return data
-
-    validated = adapter.validate_python(data)
-    return adapter.dump_python(validated, mode="json")
@@ -1,6 +1,6 @@
 # Fanout Bus Architecture

-The fanout bus is a unified system for dispatching mesh radio events (decoded messages and raw packets) to external integrations. It replaces the previous scattered singleton MQTT publishers with a modular, configurable framework.
+The fanout bus is a unified system for dispatching mesh radio events to external integrations. It replaces the previous scattered singleton MQTT publishers with a modular, configurable framework.

 ## Core Concepts

@@ -8,10 +8,15 @@ The fanout bus is a unified system for dispatching mesh radio events (decoded me
 Base class that all integration modules extend:
 - `__init__(config_id, config, *, name="")` — constructor; receives the config UUID, the type-specific config dict, and the user-assigned name
 - `start()` / `stop()` — async lifecycle (e.g. open/close connections)
- `on_message(data)` — receive decoded messages (DM/channel)
- `on_raw(data)` — receive raw RF packets
+- `on_message(data)` — receive decoded messages (scope-gated)
+- `on_raw(data)` — receive raw RF packets (scope-gated)
+- `on_contact(data)` — receive contact upserts; dispatched to all modules
+- `on_telemetry(data)` — receive repeater telemetry snapshots; dispatched to all modules
+- `on_health(data)` — receive periodic radio health snapshots; dispatched to all modules
 - `status` property (**must override**) — return `"connected"`, `"disconnected"`, or `"error"`

+All five event hooks are no-ops by default; modules override only the ones they care about.
+
 ### FanoutManager (manager.py)
 Singleton that owns all active modules and dispatches events:
 - `load_from_db()` — startup: load enabled configs, instantiate modules
@@ -19,6 +24,9 @@ Singleton that owns all active modules and dispatches events:
 - `remove_config(id)` — delete: stop and remove
 - `broadcast_message(data)` — scope-check + dispatch `on_message`
 - `broadcast_raw(data)` — scope-check + dispatch `on_raw`
+- `broadcast_contact(data)` — dispatch `on_contact` to all modules
+- `broadcast_telemetry(data)` — dispatch `on_telemetry` to all modules
+- `broadcast_health_fanout(data)` — dispatch `on_health` to all modules
 - `stop_all()` — shutdown
 - `get_statuses()` — health endpoint data

@@ -33,19 +41,65 @@ Each config has a `scope` JSON blob controlling what events reach it:
 ```
 Community MQTT always enforces `{"messages": "none", "raw_packets": "all"}`.

+Scope only gates `on_message` and `on_raw`. The `on_contact`, `on_telemetry`, and `on_health` hooks are dispatched to all modules unconditionally — modules that care about specific contacts or repeaters filter internally based on their own config.
+
 ## Event Flow

 ```
 Radio Event -> packet_processor / event_handler
-  -> broadcast_event("message"|"raw_packet", data, realtime=True)
+  -> broadcast_event("message"|"raw_packet"|"contact", data, realtime=True)
    -> WebSocket broadcast (always)
-    -> FanoutManager.broadcast_message/raw (only if realtime=True)
-      -> scope check per module
-      -> module.on_message / on_raw
+    -> FanoutManager.broadcast_message/raw/contact (only if realtime=True)
+      -> scope check per module (message/raw only)
+      -> module.on_message / on_raw / on_contact
+
+Telemetry collect (radio_sync.py / routers/repeaters.py)
+  -> RepeaterTelemetryRepository.record(...)
+  -> FanoutManager.broadcast_telemetry(data)
+    -> module.on_telemetry (all modules, unconditional)
+
+Health fanout (radio_stats.py, piggybacks on 60s stats sampling loop)
+  -> FanoutManager.broadcast_health_fanout(data)
+    -> module.on_health (all modules, unconditional)
 ```

 Setting `realtime=False` (used during historical decryption) skips fanout dispatch entirely.

+## Event Payloads
+
+### on_message(data)
+`Message.model_dump()` — the full Pydantic message model. Key fields:
+- `type` (`"PRIV"` | `"CHAN"`), `conversation_key`, `text`, `sender_name`, `sender_key`
+- `outgoing`, `acked`, `paths`, `sender_timestamp`, `received_at`
+
+### on_raw(data)
+Raw packet dict from `packet_processor.py`. Key fields:
+- `id` (storage row ID), `observation_id` (per-arrival), `raw` (hex), `timestamp`
+- `decrypted_info` (optional: `channel_key`, `contact_key`, `text`)
+
+### on_contact(data)
+`Contact.model_dump()` — the full Pydantic contact model. Key fields:
+- `public_key`, `name`, `type` (0=unknown, 1=client, 2=repeater, 3=room, 4=sensor)
+- `lat`, `lon`, `last_seen`, `first_seen`, `on_radio`
+
+### on_telemetry(data)
+Repeater telemetry snapshot, broadcast after successful `RepeaterTelemetryRepository.record()`.
+Identical shape from both auto-collect (`radio_sync.py`) and manual fetch (`routers/repeaters.py`):
+- `public_key`, `name`, `timestamp`
+- `battery_volts`, `noise_floor_dbm`, `last_rssi_dbm`, `last_snr_db`
+- `packets_received`, `packets_sent`, `airtime_seconds`, `rx_airtime_seconds`
+- `uptime_seconds`, `sent_flood`, `sent_direct`, `recv_flood`, `recv_direct`
+- `flood_dups`, `direct_dups`, `full_events`, `tx_queue_len`
+
+### on_health(data)
+Radio health + stats snapshot, broadcast every 60s by the stats sampling loop in `radio_stats.py`:
+- `connected` (bool), `connection_info` (str | None)
+- `public_key` (str | None), `name` (str | None)
+- `noise_floor_dbm`, `battery_mv`, `uptime_secs` (int | None)
+- `last_rssi` (int | None), `last_snr` (float | None)
+- `tx_air_secs`, `rx_air_secs` (int | None)
+- `packets_recv`, `packets_sent`, `flood_tx`, `direct_tx`, `flood_rx`, `direct_rx` (int | None)
+
 ## Current Module Types

 ### mqtt_private (mqtt_private.py)
@@ -38,6 +38,15 @@ class FanoutModule:
    async def on_raw(self, data: dict) -> None:
        """Called for raw RF packets. Override if needed."""

+    async def on_contact(self, data: dict) -> None:
+        """Called for contact upserts (adverts, sync). Override if needed."""
+
+    async def on_telemetry(self, data: dict) -> None:
+        """Called for repeater telemetry snapshots. Override if needed."""
+
+    async def on_health(self, data: dict) -> None:
+        """Called for periodic radio health snapshots. Override if needed."""
+
    @property
    def status(self) -> str:
        """Return 'connected', 'disconnected', or 'error'."""
@@ -164,7 +164,7 @@ class BotModule(FanoutModule):
                    ),
                    timeout=BOT_EXECUTION_TIMEOUT,
                )
-            except asyncio.TimeoutError:
+            except TimeoutError:
                logger.warning("Bot '%s' execution timed out", self.name)
                return
            except Exception:
@@ -538,7 +538,7 @@ class CommunityMqttPublisher(BaseMqttPublisher):
            self._version_event.clear()
            try:
                await asyncio.wait_for(self._version_event.wait(), timeout=30)
-            except asyncio.TimeoutError:
+            except TimeoutError:
                pass
            return False
        return True
@@ -31,12 +31,14 @@ def _register_module_types() -> None:
    from app.fanout.bot import BotModule
    from app.fanout.map_upload import MapUploadModule
    from app.fanout.mqtt_community import MqttCommunityModule
+    from app.fanout.mqtt_ha import MqttHaModule
    from app.fanout.mqtt_private import MqttPrivateModule
    from app.fanout.sqs import SqsModule
    from app.fanout.webhook import WebhookModule

    _MODULE_TYPES["mqtt_private"] = MqttPrivateModule
    _MODULE_TYPES["mqtt_community"] = MqttCommunityModule
+    _MODULE_TYPES["mqtt_ha"] = MqttHaModule
    _MODULE_TYPES["bot"] = BotModule
    _MODULE_TYPES["webhook"] = WebhookModule
    _MODULE_TYPES["apprise"] = AppriseModule
@@ -86,6 +88,11 @@ def _scope_matches_raw(scope: dict, _data: dict) -> bool:
    return scope.get("raw_packets", "none") == "all"


+def _always_match(_scope: dict, _data: dict) -> bool:
+    """Match all modules unconditionally (filtering is module-internal)."""
+    return True
+
+
 class FanoutManager:
    """Owns all active fanout modules and dispatches events."""

@@ -220,7 +227,7 @@ class FanoutManager:
            handler = getattr(module, handler_name)
            await asyncio.wait_for(handler(data), timeout=_DISPATCH_TIMEOUT_SECONDS)
            self._clear_module_error(config_id)
-        except asyncio.TimeoutError:
+        except TimeoutError:
            timeout_error = f"{handler_name} timed out after {_DISPATCH_TIMEOUT_SECONDS:.1f}s"
            self._set_module_error(config_id, timeout_error)
            logger.error(
@@ -270,6 +277,33 @@ class FanoutManager:
            log_label="on_raw",
        )

+    async def broadcast_contact(self, data: dict) -> None:
+        """Dispatch a contact upsert to all modules."""
+        await self._dispatch_matching(
+            data,
+            matcher=_always_match,
+            handler_name="on_contact",
+            log_label="on_contact",
+        )
+
+    async def broadcast_telemetry(self, data: dict) -> None:
+        """Dispatch a repeater telemetry snapshot to all modules."""
+        await self._dispatch_matching(
+            data,
+            matcher=_always_match,
+            handler_name="on_telemetry",
+            log_label="on_telemetry",
+        )
+
+    async def broadcast_health_fanout(self, data: dict) -> None:
+        """Dispatch a radio health snapshot to all modules."""
+        await self._dispatch_matching(
+            data,
+            matcher=_always_match,
+            handler_name="on_health",
+            log_label="on_health",
+        )
+
    async def stop_all(self) -> None:
        """Shutdown all modules."""
        for config_id, (module, _) in list(self._modules.items()):
@@ -144,11 +144,8 @@ class MapUploadModule(FanoutModule):
        if advert is None:
            return

-        # TODO: advert Ed25519 signature verification is skipped here.
-        # The radio has already validated the packet before passing it to RT,
-        # so re-verification is redundant in practice. If added, verify that
-        # nacl.bindings.crypto_sign_open(sig + (pubkey_bytes || timestamp_bytes),
-        # advert.public_key_bytes) succeeds before proceeding.
+        # Advert Ed25519 signature verification is intentionally skipped.
+        # The radio validates packets before passing them to RT.

        # Only process repeaters (2) and rooms (3) — any other role is rejected
        if advert.device_role not in _ALLOWED_DEVICE_ROLES:
@@ -12,6 +12,7 @@ from __future__ import annotations
 import asyncio
 import json
 import logging
+import sys
 import time
 from abc import ABC, abstractmethod
 from typing import Any
@@ -195,7 +196,7 @@ class BaseMqttPublisher(ABC):
                            self._version_event.wait(),
                            timeout=self._not_configured_timeout,
                        )
-                except asyncio.TimeoutError:
+                except TimeoutError:
                    continue
                except asyncio.CancelledError:
                    return
@@ -230,7 +231,7 @@ class BaseMqttPublisher(ABC):
                        self._version_event.clear()
                        try:
                            await asyncio.wait_for(self._version_event.wait(), timeout=60)
-                        except asyncio.TimeoutError:
+                        except TimeoutError:
                            elapsed = time.monotonic() - connect_time
                            await self._on_periodic_wake(elapsed)
                            if self._should_break_wait(elapsed):
@@ -252,6 +253,34 @@ class BaseMqttPublisher(ABC):
                self._client = None
                self._last_error = _format_error_detail(e)

+                # Windows ProactorEventLoop does not implement add_reader /
+                # add_writer, which paho-mqtt requires.  The failure can
+                # surface as a direct NotImplementedError (add_writer in
+                # __aenter__) or as a generic timeout (add_reader fails
+                # inside an event-loop callback, so paho never hears back).
+                # Either way, if we're on Windows with Proactor the root
+                # cause is the same and retrying won't help.
+                _on_proactor = (
+                    sys.platform == "win32"
+                    and type(asyncio.get_event_loop()).__name__ == "ProactorEventLoop"
+                )
+                if _on_proactor:
+                    broadcast_error(
+                        "MQTT unavailable — Windows event loop incompatible",
+                        "The default Windows event loop (ProactorEventLoop) does "
+                        "not support MQTT. Add --loop none to your uvicorn "
+                        "command and restart. See README.md for details.",
+                    )
+                    _broadcast_health()
+                    logger.error(
+                        "%s cannot run: Windows ProactorEventLoop does not "
+                        "implement add_reader/add_writer required by paho-mqtt. "
+                        "Restart uvicorn with '--loop none' to use "
+                        "SelectorEventLoop instead. Giving up (will not retry).",
+                        self._integration_label(),
+                    )
+                    return
+
                title, detail = self._on_error()
                broadcast_error(title, detail)
                _broadcast_health()
@@ -0,0 +1,757 @@
+"""Home Assistant MQTT Discovery fanout module.
+
+Publishes HA-compatible discovery configs and state updates so that mesh
+network devices appear natively in Home Assistant via its built-in MQTT
+integration. No custom HA component is needed.
+
+Entity types created:
+- Local radio: binary_sensor (connectivity) + sensors (noise floor, battery,
+  uptime, RSSI, SNR, airtime, packet counts)
+- Per tracked repeater: sensor entities for telemetry fields
+- Per tracked contact: device_tracker for GPS position
+- Messages: event entity for scope-matched messages
+"""
+
+from __future__ import annotations
+
+import logging
+import ssl
+from types import SimpleNamespace
+from typing import Any
+
+from app.fanout.base import FanoutModule, get_fanout_message_text
+from app.fanout.mqtt_base import BaseMqttPublisher
+
+logger = logging.getLogger(__name__)
+
+# ── Repeater telemetry sensor definitions ─────────────────────────────────
+
+_REPEATER_SENSORS: list[dict[str, Any]] = [
+    {
+        "field": "battery_volts",
+        "name": "Battery Voltage",
+        "object_id": "battery_voltage",
+        "device_class": "voltage",
+        "state_class": "measurement",
+        "unit": "V",
+        "precision": 2,
+    },
+    {
+        "field": "noise_floor_dbm",
+        "name": "Noise Floor",
+        "object_id": "noise_floor",
+        "device_class": "signal_strength",
+        "state_class": "measurement",
+        "unit": "dBm",
+        "precision": 0,
+    },
+    {
+        "field": "last_rssi_dbm",
+        "name": "Last RSSI",
+        "object_id": "last_rssi",
+        "device_class": "signal_strength",
+        "state_class": "measurement",
+        "unit": "dBm",
+        "precision": 0,
+    },
+    {
+        "field": "last_snr_db",
+        "name": "Last SNR",
+        "object_id": "last_snr",
+        "device_class": None,
+        "state_class": "measurement",
+        "unit": "dB",
+        "precision": 1,
+    },
+    {
+        "field": "packets_received",
+        "name": "Packets Received",
+        "object_id": "packets_received",
+        "device_class": None,
+        "state_class": "total_increasing",
+        "unit": None,
+        "precision": 0,
+    },
+    {
+        "field": "packets_sent",
+        "name": "Packets Sent",
+        "object_id": "packets_sent",
+        "device_class": None,
+        "state_class": "total_increasing",
+        "unit": None,
+        "precision": 0,
+    },
+    {
+        "field": "uptime_seconds",
+        "name": "Uptime",
+        "object_id": "uptime",
+        "device_class": "duration",
+        "state_class": None,
+        "unit": "s",
+        "precision": 0,
+    },
+]
+
+# ── LPP sensor metadata ─────────────────────────────────────────────────
+
+_LPP_HA_META: dict[str, dict[str, Any]] = {
+    "temperature": {"device_class": "temperature", "unit": "°C", "precision": 1},
+    "humidity": {"device_class": "humidity", "unit": "%", "precision": 1},
+    "barometer": {"device_class": "atmospheric_pressure", "unit": "hPa", "precision": 1},
+    "voltage": {"device_class": "voltage", "unit": "V", "precision": 2},
+    "current": {"device_class": "current", "unit": "mA", "precision": 1},
+    "luminosity": {"device_class": "illuminance", "unit": "lux", "precision": 0},
+    "power": {"device_class": "power", "unit": "W", "precision": 1},
+    "energy": {"device_class": "energy", "unit": "kWh", "precision": 2},
+    "distance": {"device_class": "distance", "unit": "mm", "precision": 0},
+    "concentration": {"device_class": None, "unit": "ppm", "precision": 0},
+    "direction": {"device_class": None, "unit": "°", "precision": 0},
+    "altitude": {"device_class": None, "unit": "m", "precision": 1},
+}
+
+
+def _lpp_sensor_key(type_name: str, channel: int) -> str:
+    """Build the flat telemetry-payload key for an LPP sensor."""
+    return f"lpp_{type_name}_ch{channel}"
+
+
+def _lpp_discovery_configs(
+    prefix: str,
+    pub_key: str,
+    device: dict,
+    lpp_sensors: list[dict],
+    state_topic: str,
+) -> list[tuple[str, dict]]:
+    """Build HA discovery configs for a repeater's LPP sensors."""
+    configs: list[tuple[str, dict]] = []
+    for sensor in lpp_sensors:
+        type_name = sensor.get("type_name", "unknown")
+        channel = sensor.get("channel", 0)
+        field = _lpp_sensor_key(type_name, channel)
+        meta = _LPP_HA_META.get(type_name, {})
+
+        nid = _node_id(pub_key)
+        object_id = field
+        display = type_name.replace("_", " ").title()
+        name = f"{display} (Ch {channel})"
+
+        cfg: dict[str, Any] = {
+            "name": name,
+            "unique_id": f"meshcore_{nid}_{object_id}",
+            "device": device,
+            "state_topic": state_topic,
+            "value_template": "{{ value_json." + field + " }}",
+            "state_class": "measurement",
+            "expire_after": 36000,
+        }
+        if meta.get("device_class"):
+            cfg["device_class"] = meta["device_class"]
+        if meta.get("unit"):
+            cfg["unit_of_measurement"] = meta["unit"]
+        if meta.get("precision") is not None:
+            cfg["suggested_display_precision"] = meta["precision"]
+
+        topic = f"homeassistant/sensor/meshcore_{nid}/{object_id}/config"
+        configs.append((topic, cfg))
+
+    return configs
+
+
+# ── Local radio sensor definitions ────────────────────────────────────────
+
+_RADIO_SENSORS: list[dict[str, Any]] = [
+    {
+        "field": "noise_floor_dbm",
+        "name": "Noise Floor",
+        "object_id": "noise_floor",
+        "device_class": "signal_strength",
+        "state_class": "measurement",
+        "unit": "dBm",
+        "precision": 0,
+    },
+    {
+        "field": "battery_volts",
+        "name": "Battery",
+        "object_id": "battery",
+        "device_class": "voltage",
+        "state_class": "measurement",
+        "unit": "V",
+        "precision": 2,
+    },
+    {
+        "field": "uptime_secs",
+        "name": "Uptime",
+        "object_id": "uptime",
+        "device_class": "duration",
+        "state_class": None,
+        "unit": "s",
+        "precision": 0,
+    },
+    {
+        "field": "last_rssi",
+        "name": "Last RSSI",
+        "object_id": "last_rssi",
+        "device_class": "signal_strength",
+        "state_class": "measurement",
+        "unit": "dBm",
+        "precision": 0,
+    },
+    {
+        "field": "last_snr",
+        "name": "Last SNR",
+        "object_id": "last_snr",
+        "device_class": None,
+        "state_class": "measurement",
+        "unit": "dB",
+        "precision": 1,
+    },
+    {
+        "field": "tx_air_secs",
+        "name": "TX Airtime",
+        "object_id": "tx_airtime",
+        "device_class": "duration",
+        "state_class": "total_increasing",
+        "unit": "s",
+        "precision": 0,
+    },
+    {
+        "field": "rx_air_secs",
+        "name": "RX Airtime",
+        "object_id": "rx_airtime",
+        "device_class": "duration",
+        "state_class": "total_increasing",
+        "unit": "s",
+        "precision": 0,
+    },
+    {
+        "field": "packets_recv",
+        "name": "Packets Received",
+        "object_id": "packets_received",
+        "device_class": None,
+        "state_class": "total_increasing",
+        "unit": None,
+        "precision": 0,
+    },
+    {
+        "field": "packets_sent",
+        "name": "Packets Sent",
+        "object_id": "packets_sent",
+        "device_class": None,
+        "state_class": "total_increasing",
+        "unit": None,
+        "precision": 0,
+    },
+]
+
+
+def _node_id(public_key: str) -> str:
+    """Derive a stable, MQTT-safe node identifier from a public key."""
+    return public_key[:12].lower()
+
+
+def _device_payload(
+    public_key: str,
+    name: str,
+    model: str,
+    *,
+    via_device_key: str | None = None,
+) -> dict[str, Any]:
+    """Build an HA device registry fragment."""
+    dev: dict[str, Any] = {
+        "identifiers": [f"meshcore_{_node_id(public_key)}"],
+        "name": name or public_key[:12],
+        "manufacturer": "MeshCore",
+        "model": model,
+    }
+    if via_device_key:
+        dev["via_device"] = f"meshcore_{_node_id(via_device_key)}"
+    return dev
+
+
+# ── MQTT publisher subclass ───────────────────────────────────────────────
+
+
+class _HaMqttPublisher(BaseMqttPublisher):
+    """Thin MQTT lifecycle wrapper for the HA discovery module."""
+
+    _backoff_max = 30
+    _log_prefix = "HA-MQTT"
+
+    def __init__(self) -> None:
+        super().__init__()
+        self._on_connected_callback: Any = None
+
+    def _is_configured(self) -> bool:
+        s = self._settings
+        return bool(s and s.broker_host)
+
+    def _build_client_kwargs(self, settings: object) -> dict[str, Any]:
+        s: Any = settings
+        kw: dict[str, Any] = {
+            "hostname": s.broker_host,
+            "port": s.broker_port,
+            "username": s.username or None,
+            "password": s.password or None,
+        }
+        if s.use_tls:
+            ctx = ssl.create_default_context()
+            if s.tls_insecure:
+                ctx.check_hostname = False
+                ctx.verify_mode = ssl.CERT_NONE
+            kw["tls_context"] = ctx
+        return kw
+
+    def _on_connected(self, settings: object) -> tuple[str, str]:
+        s: Any = settings
+        return ("HA MQTT connected", f"{s.broker_host}:{s.broker_port}")
+
+    def _on_error(self) -> tuple[str, str]:
+        return ("HA MQTT connection failure", "Please correct the settings or disable.")
+
+    async def _on_connected_async(self, settings: object) -> None:
+        if self._on_connected_callback:
+            await self._on_connected_callback()
+
+
+# ── Discovery config builders ─────────────────────────────────────────────
+
+
+def _radio_discovery_configs(
+    prefix: str,
+    radio_key: str,
+    radio_name: str,
+) -> list[tuple[str, dict]]:
+    """Build HA discovery config payloads for the local radio device."""
+    nid = _node_id(radio_key)
+    device = _device_payload(radio_key, radio_name, "Radio")
+    state_topic = f"{prefix}/{nid}/health"
+    configs: list[tuple[str, dict]] = []
+
+    # binary_sensor: connected
+    configs.append(
+        (
+            f"homeassistant/binary_sensor/meshcore_{nid}/connected/config",
+            {
+                "name": "Connected",
+                "unique_id": f"meshcore_{nid}_connected",
+                "device": device,
+                "state_topic": state_topic,
+                "value_template": "{{ 'ON' if value_json.connected else 'OFF' }}",
+                "device_class": "connectivity",
+                "payload_on": "ON",
+                "payload_off": "OFF",
+                "expire_after": 120,
+            },
+        )
+    )
+
+    # sensors from _RADIO_SENSORS (noise floor, battery, uptime, RSSI, etc.)
+    for sensor in _RADIO_SENSORS:
+        cfg: dict[str, Any] = {
+            "name": sensor["name"],
+            "unique_id": f"meshcore_{nid}_{sensor['object_id']}",
+            "device": device,
+            "state_topic": state_topic,
+            "value_template": "{{ value_json." + sensor["field"] + " }}",  # type: ignore[operator]
+            "expire_after": 120,
+        }
+        if sensor["device_class"]:
+            cfg["device_class"] = sensor["device_class"]
+        if sensor["state_class"]:
+            cfg["state_class"] = sensor["state_class"]
+        if sensor["unit"]:
+            cfg["unit_of_measurement"] = sensor["unit"]
+        if sensor.get("precision") is not None:
+            cfg["suggested_display_precision"] = sensor["precision"]
+
+        topic = f"homeassistant/sensor/meshcore_{nid}/{sensor['object_id']}/config"
+        configs.append((topic, cfg))
+
+    return configs
+
+
+def _repeater_discovery_configs(
+    prefix: str,
+    pub_key: str,
+    name: str,
+    radio_key: str | None,
+) -> list[tuple[str, dict]]:
+    """Build HA discovery config payloads for a tracked repeater."""
+    nid = _node_id(pub_key)
+    device = _device_payload(pub_key, name, "Repeater", via_device_key=radio_key)
+    state_topic = f"{prefix}/{nid}/telemetry"
+    configs: list[tuple[str, dict]] = []
+
+    for sensor in _REPEATER_SENSORS:
+        cfg: dict[str, Any] = {
+            "name": sensor["name"],
+            "unique_id": f"meshcore_{nid}_{sensor['object_id']}",
+            "device": device,
+            "state_topic": state_topic,
+            "value_template": "{{ value_json." + sensor["field"] + " }}",  # type: ignore[operator]
+        }
+        if sensor["device_class"]:
+            cfg["device_class"] = sensor["device_class"]
+        if sensor["state_class"]:
+            cfg["state_class"] = sensor["state_class"]
+        if sensor["unit"]:
+            cfg["unit_of_measurement"] = sensor["unit"]
+        if sensor.get("precision") is not None:
+            cfg["suggested_display_precision"] = sensor["precision"]
+        # 10 hours — margin over the 8-hour auto-collect cycle
+        cfg["expire_after"] = 36000
+
+        topic = f"homeassistant/sensor/meshcore_{nid}/{sensor['object_id']}/config"
+        configs.append((topic, cfg))
+
+    return configs
+
+
+def _contact_tracker_discovery_config(
+    prefix: str,
+    pub_key: str,
+    name: str,
+    radio_key: str | None,
+) -> tuple[str, dict]:
+    """Build HA discovery config for a tracked contact's device_tracker."""
+    nid = _node_id(pub_key)
+    device = _device_payload(pub_key, name, "Node", via_device_key=radio_key)
+    topic = f"homeassistant/device_tracker/meshcore_{nid}/config"
+    cfg: dict[str, Any] = {
+        "name": name or pub_key[:12],
+        "unique_id": f"meshcore_{nid}_tracker",
+        "device": device,
+        "json_attributes_topic": f"{prefix}/{nid}/gps",
+        "source_type": "gps",
+    }
+    return topic, cfg
+
+
+def _message_event_discovery_config(
+    prefix: str, radio_key: str, radio_name: str
+) -> tuple[str, dict]:
+    """Build HA discovery config for the message event entity."""
+    nid = _node_id(radio_key)
+    device = _device_payload(radio_key, radio_name, "Radio")
+    topic = f"homeassistant/event/meshcore_{nid}/messages/config"
+    cfg: dict[str, Any] = {
+        "name": "MeshCore Messages",
+        "unique_id": f"meshcore_{nid}_messages",
+        "device": device,
+        "state_topic": f"{prefix}/{nid}/events/message",
+        "event_types": ["message_received"],
+    }
+    return topic, cfg
+
+
+# ── Module class ──────────────────────────────────────────────────────────
+
+
+def _config_to_settings(config: dict) -> SimpleNamespace:
+    return SimpleNamespace(
+        broker_host=config.get("broker_host", ""),
+        broker_port=config.get("broker_port", 1883),
+        username=config.get("username", ""),
+        password=config.get("password", ""),
+        use_tls=config.get("use_tls", False),
+        tls_insecure=config.get("tls_insecure", False),
+    )
+
+
+class MqttHaModule(FanoutModule):
+    """Home Assistant MQTT Discovery fanout module."""
+
+    def __init__(self, config_id: str, config: dict, *, name: str = "") -> None:
+        super().__init__(config_id, config, name=name)
+        self._publisher = _HaMqttPublisher()
+        self._publisher.set_integration_name(name or config_id)
+        self._publisher._on_connected_callback = self._publish_discovery
+        self._discovery_topics: list[str] = []
+        self._radio_key: str | None = None
+        self._radio_name: str | None = None
+
+    @property
+    def _prefix(self) -> str:
+        return self.config.get("topic_prefix", "meshcore")
+
+    @property
+    def _tracked_contacts(self) -> list[str]:
+        return self.config.get("tracked_contacts") or []
+
+    @property
+    def _tracked_repeaters(self) -> list[str]:
+        return self.config.get("tracked_repeaters") or []
+
+    # ── Lifecycle ──────────────────────────────────────────────────────
+
+    async def start(self) -> None:
+        self._seed_radio_identity_from_runtime()
+        settings = _config_to_settings(self.config)
+        await self._publisher.start(settings)
+
+    async def stop(self) -> None:
+        await self._remove_discovery()
+        await self._publisher.stop()
+        self._discovery_topics.clear()
+
+    # ── Discovery publishing ──────────────────────────────────────────
+
+    async def _publish_discovery(self) -> None:
+        """Publish all HA discovery configs with retain=True."""
+        if not self._radio_key:
+            # Don't publish discovery until we know the radio identity —
+            # the first health heartbeat will provide it and trigger this.
+            return
+
+        configs: list[tuple[str, dict]] = []
+
+        radio_name = self._radio_name or "MeshCore Radio"
+        configs.extend(_radio_discovery_configs(self._prefix, self._radio_key, radio_name))
+
+        # Tracked repeaters — resolve names and LPP sensors from DB best-effort
+        for pub_key in self._tracked_repeaters:
+            rname = await self._resolve_contact_name(pub_key)
+            configs.extend(
+                _repeater_discovery_configs(self._prefix, pub_key, rname, self._radio_key)
+            )
+            # Dynamic LPP sensor entities from last known telemetry snapshot
+            lpp_sensors = await self._resolve_lpp_sensors(pub_key)
+            if lpp_sensors:
+                nid = _node_id(pub_key)
+                device = _device_payload(pub_key, rname, "Repeater", via_device_key=self._radio_key)
+                state_topic = f"{self._prefix}/{nid}/telemetry"
+                configs.extend(
+                    _lpp_discovery_configs(self._prefix, pub_key, device, lpp_sensors, state_topic)
+                )
+
+        # Tracked contacts — resolve names from DB best-effort
+        for pub_key in self._tracked_contacts:
+            cname = await self._resolve_contact_name(pub_key)
+            configs.append(
+                _contact_tracker_discovery_config(self._prefix, pub_key, cname, self._radio_key)
+            )
+
+        # Message event entity (namespaced to this radio)
+        configs.append(_message_event_discovery_config(self._prefix, self._radio_key, radio_name))
+
+        self._discovery_topics = [topic for topic, _ in configs]
+
+        for topic, payload in configs:
+            await self._publisher.publish(topic, payload, retain=True)
+
+        logger.info(
+            "HA MQTT: published %d discovery configs (%d repeaters, %d contacts)",
+            len(configs),
+            len(self._tracked_repeaters),
+            len(self._tracked_contacts),
+        )
+
+    async def _clear_retained_topics(self, topics: list[str]) -> None:
+        """Publish empty retained payloads to remove entries from broker."""
+        for topic in topics:
+            try:
+                if self._publisher._client:
+                    await self._publisher._client.publish(topic, b"", retain=True)
+            except Exception:
+                pass  # best-effort cleanup
+
+    async def _remove_discovery(self) -> None:
+        """Publish empty retained payloads to remove all HA entities."""
+        if not self._publisher.connected or not self._discovery_topics:
+            return
+        await self._clear_retained_topics(self._discovery_topics)
+
+    @staticmethod
+    async def _resolve_contact_name(pub_key: str) -> str:
+        """Look up a contact's display name, falling back to 12-char prefix."""
+        try:
+            from app.repository.contacts import ContactRepository
+
+            contact = await ContactRepository.get_by_key(pub_key)
+            if contact and contact.name:
+                return contact.name
+        except Exception:
+            pass
+        return pub_key[:12]
+
+    @staticmethod
+    async def _resolve_lpp_sensors(pub_key: str) -> list[dict]:
+        """Return the LPP sensor list from the most recent telemetry snapshot, or []."""
+        try:
+            from app.repository.repeater_telemetry import RepeaterTelemetryRepository
+
+            latest = await RepeaterTelemetryRepository.get_latest(pub_key)
+            if latest:
+                return latest.get("data", {}).get("lpp_sensors", [])
+        except Exception:
+            pass
+        return []
+
+    def _seed_radio_identity_from_runtime(self) -> None:
+        """Best-effort bootstrap from the currently connected radio session."""
+        try:
+            from app.services.radio_runtime import radio_runtime
+
+            if not radio_runtime.is_connected:
+                return
+
+            mc = radio_runtime.meshcore
+            self_info = mc.self_info if mc is not None else None
+            if not isinstance(self_info, dict):
+                return
+
+            pub_key = self_info.get("public_key")
+            if isinstance(pub_key, str) and pub_key.strip():
+                self._radio_key = pub_key.strip().lower()
+
+            name = self_info.get("name")
+            if isinstance(name, str) and name.strip():
+                self._radio_name = name.strip()
+        except Exception:
+            logger.debug("HA MQTT: failed to seed radio identity from runtime", exc_info=True)
+
+    # ── Event handlers ────────────────────────────────────────────────
+
+    async def on_health(self, data: dict) -> None:
+        if not self._publisher.connected:
+            return
+
+        # Cache radio identity for discovery config generation
+        pub_key = data.get("public_key")
+        if pub_key:
+            new_name = data.get("name")
+            key_changed = pub_key != self._radio_key
+            name_changed = new_name and new_name != self._radio_name
+
+            if key_changed:
+                old_key = self._radio_key
+                old_topics = list(self._discovery_topics)
+                if old_topics:
+                    await self._clear_retained_topics(old_topics)
+                    self._discovery_topics.clear()
+                self._radio_key = pub_key
+                self._radio_name = new_name
+                # Remove stale discovery entries from the old identity (e.g.
+                # "unknown" placeholder from before the radio key was known),
+                # then re-publish with the real identity.
+                if old_key is not None and not old_topics:
+                    await self._clear_retained_topics(
+                        [t for t, _ in _radio_discovery_configs(self._prefix, old_key, "")]
+                    )
+                await self._publish_discovery()
+            elif name_changed:
+                self._radio_name = new_name
+                await self._publish_discovery()
+
+        # Don't publish health state until we know the radio identity —
+        # otherwise we create a stale "unknown" device in HA.
+        if not self._radio_key:
+            return
+
+        nid = _node_id(self._radio_key)
+        payload: dict[str, Any] = {"connected": data.get("connected", False)}
+        for sensor in _RADIO_SENSORS:
+            field = sensor["field"]
+            if field is not None:
+                payload[field] = data.get(field)
+
+        # Normalize battery from millivolts to volts for consistency with
+        # repeater battery and the discovery config (unit: V, precision: 2).
+        battery_mv = data.get("battery_mv")
+        if battery_mv is not None:
+            payload["battery_volts"] = battery_mv / 1000.0
+
+        await self._publisher.publish(f"{self._prefix}/{nid}/health", payload)
+
+    async def on_contact(self, data: dict) -> None:
+        if not self._publisher.connected:
+            return
+
+        pub_key = data.get("public_key", "")
+        if pub_key not in self._tracked_contacts:
+            return
+
+        lat = data.get("lat")
+        lon = data.get("lon")
+        if lat is None or lon is None or (lat == 0.0 and lon == 0.0):
+            return
+
+        nid = _node_id(pub_key)
+        await self._publisher.publish(
+            f"{self._prefix}/{nid}/gps",
+            {
+                "latitude": lat,
+                "longitude": lon,
+                "gps_accuracy": 0,
+                "source_type": "gps",
+            },
+        )
+
+    async def on_telemetry(self, data: dict) -> None:
+        if not self._publisher.connected:
+            return
+
+        pub_key = data.get("public_key", "")
+        if pub_key not in self._tracked_repeaters:
+            return
+
+        nid = _node_id(pub_key)
+        # Publish the full telemetry dict — HA sensors use value_template
+        # to extract individual fields
+        payload: dict[str, Any] = {}
+        for s in _REPEATER_SENSORS:
+            field = s["field"]
+            if field is not None:
+                payload[field] = data.get(field)
+
+        # Flatten LPP sensors into the same payload so HA value_templates work
+        lpp_sensors: list[dict] = data.get("lpp_sensors", [])
+        rediscover = False
+        for sensor in lpp_sensors:
+            key = _lpp_sensor_key(sensor.get("type_name", "unknown"), sensor.get("channel", 0))
+            payload[key] = sensor.get("value")
+            # Check if discovery for this sensor has been published yet
+            expected_topic = f"homeassistant/sensor/meshcore_{nid}/{key}/config"
+            if expected_topic not in self._discovery_topics:
+                rediscover = True
+
+        # If new LPP sensor types appeared, re-publish discovery *before*
+        # the state payload so HA already knows the entity when the value arrives.
+        if rediscover:
+            await self._publish_discovery()
+
+        await self._publisher.publish(f"{self._prefix}/{nid}/telemetry", payload)
+
+    async def on_message(self, data: dict) -> None:
+        if not self._publisher.connected or not self._radio_key:
+            return
+
+        text = get_fanout_message_text(data)
+        nid = _node_id(self._radio_key)
+        await self._publisher.publish(
+            f"{self._prefix}/{nid}/events/message",
+            {
+                "event_type": "message_received",
+                "sender_name": data.get("sender_name", ""),
+                "sender_key": data.get("sender_key", ""),
+                "text": text,
+                "conversation_key": data.get("conversation_key", ""),
+                "message_type": data.get("type", ""),
+                "channel_name": data.get("channel_name"),
+                "outgoing": data.get("outgoing", False),
+            },
+        )
+
+    # ── Status ────────────────────────────────────────────────────────
+
+    @property
+    def status(self) -> str:
+        if not self.config.get("broker_host"):
+            return "disconnected"
+        if self._publisher.last_error:
+            return "error"
+        return "connected" if self._publisher.connected else "disconnected"
+
+    @property
+    def last_error(self) -> str | None:
+        return self._publisher.last_error
@@ -38,8 +38,17 @@ def _is_index_file(path: Path, index_file: Path) -> bool:
    return path == index_file


-def _resolve_request_origin(request: Request) -> str:
-    """Resolve the external origin, honoring common reverse-proxy headers."""
+def _resolve_request_base(request: Request) -> str:
+    """Resolve the external base URL, honoring common reverse-proxy headers.
+
+    Returns a URL like ``https://host:8000/meshcore/`` (always trailing-slash)
+    so callers can append paths directly.
+
+    Recognized headers:
+    - ``X-Forwarded-Proto`` + ``X-Forwarded-Host``: override scheme and host.
+    - ``X-Forwarded-Prefix`` (or ``X-Forwarded-Path``): sub-path prefix added
+      by the proxy (e.g. ``/meshcore``).
+    """
    forwarded_proto = request.headers.get("x-forwarded-proto")
    forwarded_host = request.headers.get("x-forwarded-host")

@@ -47,9 +56,20 @@ def _resolve_request_origin(request: Request) -> str:
        proto = forwarded_proto.split(",")[0].strip()
        host = forwarded_host.split(",")[0].strip()
        if proto and host:
-            return f"{proto}://{host}"
+            origin = f"{proto}://{host}"
+        else:
+            origin = str(request.base_url).rstrip("/")
+    else:
+        origin = str(request.base_url).rstrip("/")

-    return str(request.base_url).rstrip("/")
+    # Sub-path prefix (e.g. /meshcore) communicated by the reverse proxy
+    prefix = (
+        (request.headers.get("x-forwarded-prefix") or request.headers.get("x-forwarded-path") or "")
+        .strip()
+        .rstrip("/")
+    )
+
+    return f"{origin}{prefix}/"


 def _validate_frontend_dir(frontend_dir: Path, *, log_failures: bool = True) -> tuple[bool, Path]:
@@ -103,27 +123,27 @@ def register_frontend_static_routes(app: FastAPI, frontend_dir: Path) -> bool:

    @app.get("/site.webmanifest")
    async def serve_webmanifest(request: Request):
-        """Serve a dynamic web manifest using the active request origin."""
-        origin = _resolve_request_origin(request)
+        """Serve a dynamic web manifest using the active request base URL."""
+        base = _resolve_request_base(request)
        manifest = {
            "name": "RemoteTerm for MeshCore",
            "short_name": "RemoteTerm",
-            "id": f"{origin}/",
-            "start_url": f"{origin}/",
-            "scope": f"{origin}/",
+            "id": base,
+            "start_url": base,
+            "scope": base,
            "display": "standalone",
            "display_override": ["window-controls-overlay", "standalone", "fullscreen"],
            "theme_color": "#111419",
            "background_color": "#111419",
            "icons": [
                {
-                    "src": f"{origin}/web-app-manifest-192x192.png",
+                    "src": f"{base}web-app-manifest-192x192.png",
                    "sizes": "192x192",
                    "type": "image/png",
                    "purpose": "maskable",
                },
                {
-                    "src": f"{origin}/web-app-manifest-512x512.png",
+                    "src": f"{base}web-app-manifest-512x512.png",
                    "sizes": "512x512",
                    "type": "image/png",
                    "purpose": "maskable",
@@ -24,7 +24,7 @@ logger = logging.getLogger(__name__)

 NO_EVENT_RECEIVED_GUIDANCE = (
    "Radio command channel is unresponsive (no_event_received). Ensure that your firmware is not "
-    "incompatible, outdated, or wrong-mode (e.g. repeater, not client), and that"
+    "incompatible, outdated, or wrong-mode (e.g. repeater, not client), and that "
    "serial/TCP/BLE connectivity is successful (try another app and see if that one works?). The app cannot proceed because it cannot "
    "issue commands to the radio."
 )
@@ -1,5 +1,41 @@
-import asyncio
 import logging
+import sys
+
+# ---------------------------------------------------------------------------
+# Windows event-loop advisory for MQTT fanout
+# ---------------------------------------------------------------------------
+# On Windows, uvicorn's default event loop (ProactorEventLoop) does not
+# implement add_reader()/add_writer(), which paho-mqtt (via aiomqtt) requires.
+# We cannot fix this from inside the app — the loop is already created by the
+# time this module is imported.  Log a prominent warning so Windows operators
+# who want MQTT know to add ``--loop none`` to their uvicorn command.
+# ---------------------------------------------------------------------------
+if sys.platform == "win32":
+    import asyncio as _asyncio
+
+    _loop = _asyncio.get_event_loop()
+    _is_proactor = type(_loop).__name__ == "ProactorEventLoop"
+    if _is_proactor:
+        print(
+            "\n" + "!" * 78 + "\n"
+            "  NOTE FOR WINDOWS USERS\n" + "!" * 78 + "\n"
+            "\n"
+            "  The running event loop is ProactorEventLoop, which is not\n"
+            "  compatible with MQTT fanout (aiomqtt / paho-mqtt).\n"
+            "\n"
+            "  If you use MQTT integrations, restart with --loop none:\n"
+            "\n"
+            "    uv run uvicorn app.main:app \033[1m--loop none\033[0m"
+            " [... other options ...]\n"
+            "\n"
+            "  Everything else works fine as-is.\n"
+            "\n" + "!" * 78 + "\n",
+            file=sys.stderr,
+            flush=True,
+        )
+    del _loop, _is_proactor
+
+import asyncio
 from contextlib import asynccontextmanager
 from pathlib import Path

@@ -21,6 +57,7 @@ from app.radio_sync import (
    stop_message_polling,
    stop_periodic_advert,
    stop_periodic_sync,
+    stop_telemetry_collect,
 )
 from app.routers import (
    channels,
@@ -39,8 +76,8 @@ from app.routers import (
    ws,
 )
 from app.security import add_optional_basic_auth_middleware
-from app.services.radio_noise_floor import start_noise_floor_sampling, stop_noise_floor_sampling
 from app.services.radio_runtime import radio_runtime as radio_manager
+from app.services.radio_stats import start_radio_stats_sampling, stop_radio_stats_sampling
 from app.version_info import get_app_build_info

 setup_logging()
@@ -71,7 +108,7 @@ async def lifespan(app: FastAPI):
    from app.radio_sync import ensure_default_channels

    await ensure_default_channels()
-    await start_noise_floor_sampling()
+    await start_radio_stats_sampling()

    # Always start connection monitor (even if initial connection failed)
    await radio_manager.start_connection_monitor()
@@ -100,9 +137,10 @@ async def lifespan(app: FastAPI):
    await radio_manager.stop_connection_monitor()
    await stop_background_contact_reconciliation()
    await stop_message_polling()
-    await stop_noise_floor_sampling()
+    await stop_radio_stats_sampling()
    await stop_periodic_advert()
    await stop_periodic_sync()
+    await stop_telemetry_collect()
    if radio_manager.meshcore:
        await radio_manager.meshcore.stop_auto_message_fetching()
    await radio_manager.disconnect()
@@ -0,0 +1,38 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add last_read_at column to contacts and channels tables.
+
+    This enables server-side read state tracking, replacing the localStorage
+    approach for consistent read state across devices.
+
+    ALTER TABLE ADD COLUMN is safe - it preserves existing data and handles
+    the "column already exists" case gracefully.
+    """
+    # Add to contacts table
+    try:
+        await conn.execute("ALTER TABLE contacts ADD COLUMN last_read_at INTEGER")
+        logger.debug("Added last_read_at to contacts table")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column name" in str(e).lower():
+            logger.debug("contacts.last_read_at already exists, skipping")
+        else:
+            raise
+
+    # Add to channels table
+    try:
+        await conn.execute("ALTER TABLE channels ADD COLUMN last_read_at INTEGER")
+        logger.debug("Added last_read_at to channels table")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column name" in str(e).lower():
+            logger.debug("channels.last_read_at already exists, skipping")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,32 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Drop unused decrypt_attempts and last_attempt columns from raw_packets.
+
+    These columns were added for a retry-limiting feature that was never implemented.
+    They are written to but never read, so we can safely remove them.
+
+    SQLite 3.35.0+ supports ALTER TABLE DROP COLUMN. For older versions,
+    we silently skip (the columns will remain but are harmless).
+    """
+    for column in ["decrypt_attempts", "last_attempt"]:
+        try:
+            await conn.execute(f"ALTER TABLE raw_packets DROP COLUMN {column}")
+            logger.debug("Dropped %s from raw_packets table", column)
+        except aiosqlite.OperationalError as e:
+            error_msg = str(e).lower()
+            if "no such column" in error_msg:
+                logger.debug("raw_packets.%s already dropped, skipping", column)
+            elif "syntax error" in error_msg or "drop column" in error_msg:
+                # SQLite version doesn't support DROP COLUMN - harmless, column stays
+                logger.debug("SQLite doesn't support DROP COLUMN, %s column will remain", column)
+            else:
+                raise
+
+    await conn.commit()
@@ -0,0 +1,49 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Drop the decrypted column and update indexes.
+
+    The decrypted column is redundant with message_id - a packet is decrypted
+    iff message_id IS NOT NULL. We replace the decrypted index with a message_id index.
+
+    SQLite 3.35.0+ supports ALTER TABLE DROP COLUMN. For older versions,
+    we silently skip the column drop but still update the index.
+    """
+    # First, drop the old index on decrypted (safe even if it doesn't exist)
+    try:
+        await conn.execute("DROP INDEX IF EXISTS idx_raw_packets_decrypted")
+        logger.debug("Dropped idx_raw_packets_decrypted index")
+    except aiosqlite.OperationalError:
+        pass  # Index didn't exist
+
+    # Create new index on message_id for efficient undecrypted packet queries
+    try:
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_raw_packets_message_id ON raw_packets(message_id)"
+        )
+        logger.debug("Created idx_raw_packets_message_id index")
+    except aiosqlite.OperationalError as e:
+        if "already exists" not in str(e).lower():
+            raise
+
+    # Try to drop the decrypted column
+    try:
+        await conn.execute("ALTER TABLE raw_packets DROP COLUMN decrypted")
+        logger.debug("Dropped decrypted from raw_packets table")
+    except aiosqlite.OperationalError as e:
+        error_msg = str(e).lower()
+        if "no such column" in error_msg:
+            logger.debug("raw_packets.decrypted already dropped, skipping")
+        elif "syntax error" in error_msg or "drop column" in error_msg:
+            # SQLite version doesn't support DROP COLUMN - harmless, column stays
+            logger.debug("SQLite doesn't support DROP COLUMN, decrypted column will remain")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,24 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add payload_hash column to raw_packets for deduplication.
+
+    This column stores the SHA-256 hash of the packet payload (excluding routing/path info).
+    It will be used with a unique index to prevent duplicate packets from being stored.
+    """
+    try:
+        await conn.execute("ALTER TABLE raw_packets ADD COLUMN payload_hash TEXT")
+        logger.debug("Added payload_hash column to raw_packets table")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column name" in str(e).lower():
+            logger.debug("raw_packets.payload_hash already exists, skipping")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,126 @@
+import logging
+from hashlib import sha256
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+def _extract_payload_for_hash(raw_packet: bytes) -> bytes | None:
+    """
+    Extract payload from a raw packet for hashing using canonical framing validation.
+
+    Returns the payload bytes, or None if packet is malformed.
+    """
+    from app.path_utils import parse_packet_envelope
+
+    envelope = parse_packet_envelope(raw_packet)
+    return envelope.payload if envelope is not None else None
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Backfill payload_hash for existing packets and remove duplicates.
+
+    This may take a while for large databases. Progress is logged.
+    After backfilling, a unique index is created to prevent future duplicates.
+    """
+    # Get count first
+    cursor = await conn.execute("SELECT COUNT(*) FROM raw_packets WHERE payload_hash IS NULL")
+    row = await cursor.fetchone()
+    total = row[0] if row else 0
+
+    if total == 0:
+        logger.debug("No packets need hash backfill")
+    else:
+        logger.info("Backfilling payload hashes for %d packets. This may take a while...", total)
+
+        # Process in batches to avoid memory issues
+        batch_size = 1000
+        processed = 0
+        duplicates_deleted = 0
+
+        # Track seen hashes to identify duplicates (keep oldest = lowest ID)
+        seen_hashes: dict[str, int] = {}  # hash -> oldest packet ID
+
+        # First pass: compute hashes and identify duplicates
+        cursor = await conn.execute("SELECT id, data FROM raw_packets ORDER BY id ASC")
+
+        packets_to_update: list[tuple[str, int]] = []  # (hash, id)
+        ids_to_delete: list[int] = []
+
+        while True:
+            rows = await cursor.fetchmany(batch_size)
+            if not rows:
+                break
+
+            for row in rows:
+                packet_id = row[0]
+                packet_data = bytes(row[1])
+
+                # Extract payload and compute hash
+                payload = _extract_payload_for_hash(packet_data)
+                if payload:
+                    payload_hash = sha256(payload).hexdigest()
+                else:
+                    # For malformed packets, hash the full data
+                    payload_hash = sha256(packet_data).hexdigest()
+
+                if payload_hash in seen_hashes:
+                    # Duplicate - mark for deletion (we keep the older one)
+                    ids_to_delete.append(packet_id)
+                    duplicates_deleted += 1
+                else:
+                    # New hash - keep this packet
+                    seen_hashes[payload_hash] = packet_id
+                    packets_to_update.append((payload_hash, packet_id))
+
+                processed += 1
+
+            if processed % 10000 == 0:
+                logger.info("Processed %d/%d packets...", processed, total)
+
+        # Second pass: update hashes for packets we're keeping
+        total_updates = len(packets_to_update)
+        logger.info("Updating %d packets with hashes...", total_updates)
+        for idx, (payload_hash, packet_id) in enumerate(packets_to_update, 1):
+            await conn.execute(
+                "UPDATE raw_packets SET payload_hash = ? WHERE id = ?",
+                (payload_hash, packet_id),
+            )
+            if idx % 10000 == 0:
+                logger.info("Updated %d/%d packets...", idx, total_updates)
+
+        # Third pass: delete duplicates
+        if ids_to_delete:
+            total_deletes = len(ids_to_delete)
+            logger.info("Removing %d duplicate packets...", total_deletes)
+            deleted_count = 0
+            # Delete in batches to avoid "too many SQL variables" error
+            for i in range(0, len(ids_to_delete), 500):
+                batch = ids_to_delete[i : i + 500]
+                placeholders = ",".join("?" * len(batch))
+                await conn.execute(f"DELETE FROM raw_packets WHERE id IN ({placeholders})", batch)
+                deleted_count += len(batch)
+                if deleted_count % 10000 < 500:  # Log roughly every 10k
+                    logger.info("Removed %d/%d duplicates...", deleted_count, total_deletes)
+
+        await conn.commit()
+        logger.info(
+            "Hash backfill complete: %d packets updated, %d duplicates removed",
+            len(packets_to_update),
+            duplicates_deleted,
+        )
+
+    # Create unique index on payload_hash (this enforces uniqueness going forward)
+    try:
+        await conn.execute(
+            "CREATE UNIQUE INDEX IF NOT EXISTS idx_raw_packets_payload_hash "
+            "ON raw_packets(payload_hash)"
+        )
+        logger.debug("Created unique index on payload_hash")
+    except aiosqlite.OperationalError as e:
+        if "already exists" not in str(e).lower():
+            raise
+
+    await conn.commit()
@@ -0,0 +1,42 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Replace path_len INTEGER column with path TEXT column in messages table.
+
+    The path column stores the hex-encoded routing path bytes. Path length can
+    be derived from the hex string (2 chars per byte = 1 hop).
+
+    SQLite 3.35.0+ supports ALTER TABLE DROP COLUMN. For older versions,
+    we silently skip the drop (the column will remain but is unused).
+    """
+    # First, add the new path column
+    try:
+        await conn.execute("ALTER TABLE messages ADD COLUMN path TEXT")
+        logger.debug("Added path column to messages table")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column name" in str(e).lower():
+            logger.debug("messages.path already exists, skipping")
+        else:
+            raise
+
+    # Try to drop the old path_len column
+    try:
+        await conn.execute("ALTER TABLE messages DROP COLUMN path_len")
+        logger.debug("Dropped path_len from messages table")
+    except aiosqlite.OperationalError as e:
+        error_msg = str(e).lower()
+        if "no such column" in error_msg:
+            logger.debug("messages.path_len already dropped, skipping")
+        elif "syntax error" in error_msg or "drop column" in error_msg:
+            # SQLite version doesn't support DROP COLUMN - harmless, column stays
+            logger.debug("SQLite doesn't support DROP COLUMN, path_len column will remain")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,96 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+def _extract_path_from_packet(raw_packet: bytes) -> str | None:
+    """
+    Extract path hex string from a raw packet using canonical framing validation.
+
+    Returns the path as a hex string, or None if packet is malformed.
+    """
+    from app.path_utils import parse_packet_envelope
+
+    envelope = parse_packet_envelope(raw_packet)
+    return envelope.path.hex() if envelope is not None else None
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Backfill path column for messages that have linked raw_packets.
+
+    For each message with a linked raw_packet (via message_id), extract the
+    path from the raw packet and update the message.
+
+    Only updates incoming messages (outgoing=0) since outgoing messages
+    don't have meaningful path data.
+    """
+    # Get count of messages that need backfill
+    cursor = await conn.execute(
+        """
+        SELECT COUNT(*)
+        FROM messages m
+        JOIN raw_packets rp ON rp.message_id = m.id
+        WHERE m.path IS NULL AND m.outgoing = 0
+        """
+    )
+    row = await cursor.fetchone()
+    total = row[0] if row else 0
+
+    if total == 0:
+        logger.debug("No messages need path backfill")
+        return
+
+    logger.info("Backfilling path for %d messages. This may take a while...", total)
+
+    # Process in batches
+    batch_size = 1000
+    processed = 0
+    updated = 0
+
+    cursor = await conn.execute(
+        """
+        SELECT m.id, rp.data
+        FROM messages m
+        JOIN raw_packets rp ON rp.message_id = m.id
+        WHERE m.path IS NULL AND m.outgoing = 0
+        ORDER BY m.id ASC
+        """
+    )
+
+    updates: list[tuple[str, int]] = []  # (path, message_id)
+
+    while True:
+        rows = await cursor.fetchmany(batch_size)
+        if not rows:
+            break
+
+        for row in rows:
+            message_id = row[0]
+            packet_data = bytes(row[1])
+
+            path_hex = _extract_path_from_packet(packet_data)
+            if path_hex is not None:
+                updates.append((path_hex, message_id))
+
+            processed += 1
+
+        if processed % 10000 == 0:
+            logger.info("Processed %d/%d messages...", processed, total)
+
+    # Apply updates in batches
+    if updates:
+        logger.info("Updating %d messages with path data...", len(updates))
+        for idx, (path_hex, message_id) in enumerate(updates, 1):
+            await conn.execute(
+                "UPDATE messages SET path = ? WHERE id = ?",
+                (path_hex, message_id),
+            )
+            updated += 1
+            if idx % 10000 == 0:
+                logger.info("Updated %d/%d messages...", idx, len(updates))
+
+    await conn.commit()
+    logger.info("Path backfill complete: %d messages updated", updated)
@@ -0,0 +1,66 @@
+import json
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Convert path TEXT column to paths TEXT column storing JSON array.
+
+    The new format stores multiple paths as a JSON array of objects:
+    [{"path": "1A2B", "received_at": 1234567890}, ...]
+
+    This enables tracking multiple delivery paths for the same message
+    (e.g., when a message is received via different repeater routes).
+    """
+
+    # First, add the new paths column
+    try:
+        await conn.execute("ALTER TABLE messages ADD COLUMN paths TEXT")
+        logger.debug("Added paths column to messages table")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column name" in str(e).lower():
+            logger.debug("messages.paths already exists, skipping column add")
+        else:
+            raise
+
+    # Migrate existing path data to paths array format
+    cursor = await conn.execute(
+        "SELECT id, path, received_at FROM messages WHERE path IS NOT NULL AND paths IS NULL"
+    )
+    rows = list(await cursor.fetchall())
+
+    if rows:
+        logger.info("Converting %d messages from path to paths array format...", len(rows))
+        for row in rows:
+            message_id = row[0]
+            old_path = row[1]
+            received_at = row[2]
+
+            # Convert single path to array format
+            paths_json = json.dumps([{"path": old_path, "received_at": received_at}])
+            await conn.execute(
+                "UPDATE messages SET paths = ? WHERE id = ?",
+                (paths_json, message_id),
+            )
+
+        logger.info("Converted %d messages to paths array format", len(rows))
+
+    # Try to drop the old path column (SQLite 3.35.0+ only)
+    try:
+        await conn.execute("ALTER TABLE messages DROP COLUMN path")
+        logger.debug("Dropped path column from messages table")
+    except aiosqlite.OperationalError as e:
+        error_msg = str(e).lower()
+        if "no such column" in error_msg:
+            logger.debug("messages.path already dropped, skipping")
+        elif "syntax error" in error_msg or "drop column" in error_msg:
+            # SQLite version doesn't support DROP COLUMN - harmless, column stays
+            logger.debug("SQLite doesn't support DROP COLUMN, path column will remain")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,41 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Create app_settings table for persistent application preferences.
+
+    This table stores:
+    - max_radio_contacts: Configured radio contact capacity baseline for maintenance thresholds
+    - favorites: JSON array of favorite conversations [{type, id}, ...]
+    - auto_decrypt_dm_on_advert: Whether to attempt historical DM decryption on new contact
+    - sidebar_sort_order: 'recent' or 'alpha' for sidebar sorting
+    - last_message_times: JSON object mapping conversation keys to timestamps
+    - preferences_migrated: Flag to track if localStorage has been migrated
+
+    The table uses a single-row pattern (id=1) for simplicity.
+    """
+    await conn.execute(
+        """
+        CREATE TABLE IF NOT EXISTS app_settings (
+            id INTEGER PRIMARY KEY CHECK (id = 1),
+            max_radio_contacts INTEGER DEFAULT 200,
+            favorites TEXT DEFAULT '[]',
+            auto_decrypt_dm_on_advert INTEGER DEFAULT 1,
+            sidebar_sort_order TEXT DEFAULT 'recent',
+            last_message_times TEXT DEFAULT '{}',
+            preferences_migrated INTEGER DEFAULT 0
+        )
+        """
+    )
+
+    # Initialize with default row (use only the id column so this works
+    # regardless of which columns exist — defaults fill the rest).
+    await conn.execute("INSERT OR IGNORE INTO app_settings (id) VALUES (1)")
+
+    await conn.commit()
+    logger.debug("Created app_settings table with default values")
@@ -0,0 +1,23 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add advert_interval column to app_settings table.
+
+    This enables configurable periodic advertisement interval (default 0 = disabled).
+    """
+    try:
+        await conn.execute("ALTER TABLE app_settings ADD COLUMN advert_interval INTEGER DEFAULT 0")
+        logger.debug("Added advert_interval column to app_settings")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column" in str(e).lower():
+            logger.debug("advert_interval column already exists, skipping")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,24 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add last_advert_time column to app_settings table.
+
+    This tracks when the last advertisement was sent, ensuring we never
+    advertise faster than the configured advert_interval.
+    """
+    try:
+        await conn.execute("ALTER TABLE app_settings ADD COLUMN last_advert_time INTEGER DEFAULT 0")
+        logger.debug("Added last_advert_time column to app_settings")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column" in str(e).lower():
+            logger.debug("last_advert_time column already exists, skipping")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,33 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add bot_enabled and bot_code columns to app_settings table.
+
+    This enables user-defined Python code to be executed when messages are received,
+    allowing for custom bot responses.
+    """
+    try:
+        await conn.execute("ALTER TABLE app_settings ADD COLUMN bot_enabled INTEGER DEFAULT 0")
+        logger.debug("Added bot_enabled column to app_settings")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column" in str(e).lower():
+            logger.debug("bot_enabled column already exists, skipping")
+        else:
+            raise
+
+    try:
+        await conn.execute("ALTER TABLE app_settings ADD COLUMN bot_code TEXT DEFAULT ''")
+        logger.debug("Added bot_code column to app_settings")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column" in str(e).lower():
+            logger.debug("bot_code column already exists, skipping")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,76 @@
+import json
+import logging
+import uuid
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Convert single bot_enabled/bot_code to multi-bot format.
+
+    Adds a 'bots' TEXT column storing a JSON array of bot configs:
+    [{"id": "uuid", "name": "Bot 1", "enabled": true, "code": "..."}]
+
+    If existing bot_code is non-empty OR bot_enabled is true, migrates
+    to a single bot named "Bot 1". Otherwise, creates empty array.
+
+    Attempts to drop the old bot_enabled and bot_code columns.
+    """
+
+    # Add new bots column
+    try:
+        await conn.execute("ALTER TABLE app_settings ADD COLUMN bots TEXT DEFAULT '[]'")
+        logger.debug("Added bots column to app_settings")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column" in str(e).lower():
+            logger.debug("bots column already exists, skipping")
+        else:
+            raise
+
+    # Migrate existing bot data
+    cursor = await conn.execute("SELECT bot_enabled, bot_code FROM app_settings WHERE id = 1")
+    row = await cursor.fetchone()
+
+    if row:
+        bot_enabled = bool(row[0]) if row[0] is not None else False
+        bot_code = row[1] or ""
+
+        # If there's existing bot data, migrate it
+        if bot_code.strip() or bot_enabled:
+            bots = [
+                {
+                    "id": str(uuid.uuid4()),
+                    "name": "Bot 1",
+                    "enabled": bot_enabled,
+                    "code": bot_code,
+                }
+            ]
+            bots_json = json.dumps(bots)
+            logger.info("Migrating existing bot to multi-bot format: enabled=%s", bot_enabled)
+        else:
+            bots_json = "[]"
+
+        await conn.execute(
+            "UPDATE app_settings SET bots = ? WHERE id = 1",
+            (bots_json,),
+        )
+
+    # Try to drop old columns (SQLite 3.35.0+ only)
+    for column in ["bot_enabled", "bot_code"]:
+        try:
+            await conn.execute(f"ALTER TABLE app_settings DROP COLUMN {column}")
+            logger.debug("Dropped %s column from app_settings", column)
+        except aiosqlite.OperationalError as e:
+            error_msg = str(e).lower()
+            if "no such column" in error_msg:
+                logger.debug("app_settings.%s already dropped, skipping", column)
+            elif "syntax error" in error_msg or "drop column" in error_msg:
+                # SQLite version doesn't support DROP COLUMN - harmless, column stays
+                logger.debug("SQLite doesn't support DROP COLUMN, %s column will remain", column)
+            else:
+                raise
+
+    await conn.commit()
@@ -0,0 +1,152 @@
+import json
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Lowercase all contact public keys and related data for case-insensitive matching.
+
+    Updates:
+    - contacts.public_key (PRIMARY KEY) via temp table swap
+    - messages.conversation_key for PRIV messages
+    - app_settings.favorites (contact IDs)
+    - app_settings.last_message_times (contact- prefixed keys)
+
+    Handles case collisions by keeping the most-recently-seen contact.
+    """
+
+    # 1. Lowercase message conversation keys for private messages
+    try:
+        await conn.execute(
+            "UPDATE messages SET conversation_key = lower(conversation_key) WHERE type = 'PRIV'"
+        )
+        logger.debug("Lowercased PRIV message conversation_keys")
+    except aiosqlite.OperationalError as e:
+        if "no such table" in str(e).lower():
+            logger.debug("messages table does not exist yet, skipping conversation_key lowercase")
+        else:
+            raise
+
+    # 2. Check if contacts table exists before proceeding
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='contacts'"
+    )
+    if not await cursor.fetchone():
+        logger.debug("contacts table does not exist yet, skipping key lowercase")
+        await conn.commit()
+        return
+
+    # 3. Handle contacts table - check for case collisions first
+    cursor = await conn.execute(
+        "SELECT lower(public_key) as lk, COUNT(*) as cnt "
+        "FROM contacts GROUP BY lower(public_key) HAVING COUNT(*) > 1"
+    )
+    collisions = list(await cursor.fetchall())
+
+    if collisions:
+        logger.warning(
+            "Found %d case-colliding contact groups, keeping most-recently-seen",
+            len(collisions),
+        )
+        for row in collisions:
+            lower_key = row[0]
+            # Delete all but the most recently seen
+            await conn.execute(
+                """DELETE FROM contacts WHERE public_key IN (
+                    SELECT public_key FROM contacts
+                    WHERE lower(public_key) = ?
+                    ORDER BY COALESCE(last_seen, 0) DESC
+                    LIMIT -1 OFFSET 1
+                )""",
+                (lower_key,),
+            )
+
+    # 3. Rebuild contacts with lowercased keys
+    # Get the actual column names from the table (handles different schema versions)
+    cursor = await conn.execute("PRAGMA table_info(contacts)")
+    columns_info = await cursor.fetchall()
+    all_columns = [col[1] for col in columns_info]  # col[1] is column name
+
+    # Build column lists, lowering public_key
+    select_cols = ", ".join(f"lower({c})" if c == "public_key" else c for c in all_columns)
+    col_defs = []
+    for col in columns_info:
+        name, col_type, _notnull, default, pk = col[1], col[2], col[3], col[4], col[5]
+        parts = [name, col_type or "TEXT"]
+        if pk:
+            parts.append("PRIMARY KEY")
+        if default is not None:
+            parts.append(f"DEFAULT {default}")
+        col_defs.append(" ".join(parts))
+
+    create_sql = f"CREATE TABLE contacts_new ({', '.join(col_defs)})"
+    await conn.execute(create_sql)
+    await conn.execute(f"INSERT INTO contacts_new SELECT {select_cols} FROM contacts")
+    await conn.execute("DROP TABLE contacts")
+    await conn.execute("ALTER TABLE contacts_new RENAME TO contacts")
+
+    # Recreate the on_radio index (if column exists)
+    if "on_radio" in all_columns:
+        await conn.execute("CREATE INDEX IF NOT EXISTS idx_contacts_on_radio ON contacts(on_radio)")
+
+    # 4. Lowercase contact IDs in favorites JSON (if app_settings exists)
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='app_settings'"
+    )
+    if not await cursor.fetchone():
+        await conn.commit()
+        logger.info("Lowercased all contact public keys (no app_settings table)")
+        return
+
+    cursor = await conn.execute("SELECT favorites FROM app_settings WHERE id = 1")
+    row = await cursor.fetchone()
+    if row and row[0]:
+        try:
+            favorites = json.loads(row[0])
+            updated = False
+            for fav in favorites:
+                if fav.get("type") == "contact" and fav.get("id"):
+                    new_id = fav["id"].lower()
+                    if new_id != fav["id"]:
+                        fav["id"] = new_id
+                        updated = True
+            if updated:
+                await conn.execute(
+                    "UPDATE app_settings SET favorites = ? WHERE id = 1",
+                    (json.dumps(favorites),),
+                )
+                logger.debug("Lowercased contact IDs in favorites")
+        except (json.JSONDecodeError, TypeError):
+            pass
+
+    # 5. Lowercase contact keys in last_message_times JSON
+    cursor = await conn.execute("SELECT last_message_times FROM app_settings WHERE id = 1")
+    row = await cursor.fetchone()
+    if row and row[0]:
+        try:
+            times = json.loads(row[0])
+            new_times = {}
+            updated = False
+            for key, val in times.items():
+                if key.startswith("contact-"):
+                    new_key = "contact-" + key[8:].lower()
+                    if new_key != key:
+                        updated = True
+                    new_times[new_key] = val
+                else:
+                    new_times[key] = val
+            if updated:
+                await conn.execute(
+                    "UPDATE app_settings SET last_message_times = ? WHERE id = 1",
+                    (json.dumps(new_times),),
+                )
+                logger.debug("Lowercased contact keys in last_message_times")
+        except (json.JSONDecodeError, TypeError):
+            pass
+
+    await conn.commit()
+    logger.info("Lowercased all contact public keys")
@@ -0,0 +1,44 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Fix NULL sender_timestamp values and add null-safe dedup index.
+
+    1. Set sender_timestamp = received_at for any messages with NULL sender_timestamp
+    2. Create a null-safe unique index as belt-and-suspenders protection
+    """
+    # Check if messages table exists
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='messages'"
+    )
+    if not await cursor.fetchone():
+        logger.debug("messages table does not exist yet, skipping NULL sender_timestamp fix")
+        await conn.commit()
+        return
+
+    # Backfill NULL sender_timestamps with received_at
+    cursor = await conn.execute(
+        "UPDATE messages SET sender_timestamp = received_at WHERE sender_timestamp IS NULL"
+    )
+    if cursor.rowcount > 0:
+        logger.info("Backfilled %d messages with NULL sender_timestamp", cursor.rowcount)
+
+    # Try to create null-safe dedup index (may fail if existing duplicates exist)
+    try:
+        await conn.execute(
+            """CREATE UNIQUE INDEX IF NOT EXISTS idx_messages_dedup_null_safe
+               ON messages(type, conversation_key, text, COALESCE(sender_timestamp, 0))"""
+        )
+        logger.debug("Created null-safe dedup index")
+    except aiosqlite.IntegrityError:
+        logger.warning(
+            "Could not create null-safe dedup index due to existing duplicates - "
+            "the application-level dedup will handle these"
+        )
+
+    await conn.commit()
@@ -0,0 +1,26 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add experimental_channel_double_send column to app_settings table.
+
+    When enabled, channel sends perform an immediate byte-perfect duplicate send
+    using the same timestamp bytes.
+    """
+    try:
+        await conn.execute(
+            "ALTER TABLE app_settings ADD COLUMN experimental_channel_double_send INTEGER DEFAULT 0"
+        )
+        logger.debug("Added experimental_channel_double_send column to app_settings")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column" in str(e).lower():
+            logger.debug("experimental_channel_double_send column already exists, skipping")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,31 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Drop experimental_channel_double_send column from app_settings.
+
+    This feature is replaced by a user-triggered resend button.
+    SQLite 3.35.0+ supports ALTER TABLE DROP COLUMN. For older versions,
+    we silently skip (the column will remain but is unused).
+    """
+    try:
+        await conn.execute("ALTER TABLE app_settings DROP COLUMN experimental_channel_double_send")
+        logger.debug("Dropped experimental_channel_double_send from app_settings")
+    except aiosqlite.OperationalError as e:
+        error_msg = str(e).lower()
+        if "no such column" in error_msg:
+            logger.debug("app_settings.experimental_channel_double_send already dropped, skipping")
+        elif "syntax error" in error_msg or "drop column" in error_msg:
+            logger.debug(
+                "SQLite doesn't support DROP COLUMN, "
+                "experimental_channel_double_send column will remain"
+            )
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,64 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Drop the UNIQUE constraint on raw_packets.data via table rebuild.
+
+    This constraint creates a large autoindex (~30 MB on a 340K-row database) that
+    stores a complete copy of every raw packet BLOB in a B-tree. Deduplication is
+    already handled by the unique index on payload_hash, making the data UNIQUE
+    constraint pure storage overhead.
+
+    Requires table recreation since SQLite doesn't support DROP CONSTRAINT.
+    """
+    # Check if the autoindex exists (indicates UNIQUE constraint on data)
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='index' "
+        "AND name='sqlite_autoindex_raw_packets_1'"
+    )
+    if not await cursor.fetchone():
+        logger.debug("raw_packets.data UNIQUE constraint already absent, skipping rebuild")
+        await conn.commit()
+        return
+
+    logger.info("Rebuilding raw_packets table to remove UNIQUE(data) constraint...")
+
+    # Get current columns from the existing table
+    cursor = await conn.execute("PRAGMA table_info(raw_packets)")
+    old_cols = {col[1] for col in await cursor.fetchall()}
+
+    # Target schema without UNIQUE on data
+    await conn.execute("""
+        CREATE TABLE raw_packets_new (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            timestamp INTEGER NOT NULL,
+            data BLOB NOT NULL,
+            message_id INTEGER,
+            payload_hash TEXT,
+            FOREIGN KEY (message_id) REFERENCES messages(id)
+        )
+    """)
+
+    # Copy only columns that exist in both old and new tables
+    new_cols = {"id", "timestamp", "data", "message_id", "payload_hash"}
+    copy_cols = ", ".join(sorted(c for c in new_cols if c in old_cols))
+
+    await conn.execute(
+        f"INSERT INTO raw_packets_new ({copy_cols}) SELECT {copy_cols} FROM raw_packets"
+    )
+    await conn.execute("DROP TABLE raw_packets")
+    await conn.execute("ALTER TABLE raw_packets_new RENAME TO raw_packets")
+
+    # Recreate indexes
+    await conn.execute(
+        "CREATE UNIQUE INDEX idx_raw_packets_payload_hash ON raw_packets(payload_hash)"
+    )
+    await conn.execute("CREATE INDEX idx_raw_packets_message_id ON raw_packets(message_id)")
+
+    await conn.commit()
+    logger.info("raw_packets table rebuilt without UNIQUE(data) constraint")
@@ -0,0 +1,83 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Drop the UNIQUE(type, conversation_key, text, sender_timestamp) constraint on messages.
+
+    This constraint creates a large autoindex (~13 MB on a 112K-row database) that
+    stores the full message text in a B-tree. The idx_messages_dedup_null_safe unique
+    index already provides identical dedup protection — no rows have NULL
+    sender_timestamp since migration 15 backfilled them all.
+
+    INSERT OR IGNORE still works correctly because it checks all unique constraints,
+    including unique indexes like idx_messages_dedup_null_safe.
+
+    Requires table recreation since SQLite doesn't support DROP CONSTRAINT.
+    """
+    # Check if the autoindex exists (indicates UNIQUE constraint)
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='index' AND name='sqlite_autoindex_messages_1'"
+    )
+    if not await cursor.fetchone():
+        logger.debug("messages UNIQUE constraint already absent, skipping rebuild")
+        await conn.commit()
+        return
+
+    logger.info("Rebuilding messages table to remove UNIQUE constraint...")
+
+    # Get current columns from the existing table
+    cursor = await conn.execute("PRAGMA table_info(messages)")
+    old_cols = {col[1] for col in await cursor.fetchall()}
+
+    # Target schema without the UNIQUE table constraint
+    await conn.execute("""
+        CREATE TABLE messages_new (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            type TEXT NOT NULL,
+            conversation_key TEXT NOT NULL,
+            text TEXT NOT NULL,
+            sender_timestamp INTEGER,
+            received_at INTEGER NOT NULL,
+            txt_type INTEGER DEFAULT 0,
+            signature TEXT,
+            outgoing INTEGER DEFAULT 0,
+            acked INTEGER DEFAULT 0,
+            paths TEXT
+        )
+    """)
+
+    # Copy only columns that exist in both old and new tables
+    new_cols = {
+        "id",
+        "type",
+        "conversation_key",
+        "text",
+        "sender_timestamp",
+        "received_at",
+        "txt_type",
+        "signature",
+        "outgoing",
+        "acked",
+        "paths",
+    }
+    copy_cols = ", ".join(sorted(c for c in new_cols if c in old_cols))
+
+    await conn.execute(f"INSERT INTO messages_new ({copy_cols}) SELECT {copy_cols} FROM messages")
+    await conn.execute("DROP TABLE messages")
+    await conn.execute("ALTER TABLE messages_new RENAME TO messages")
+
+    # Recreate indexes
+    await conn.execute("CREATE INDEX idx_messages_conversation ON messages(type, conversation_key)")
+    await conn.execute("CREATE INDEX idx_messages_received ON messages(received_at)")
+    await conn.execute(
+        """CREATE UNIQUE INDEX idx_messages_dedup_null_safe
+           ON messages(type, conversation_key, text, COALESCE(sender_timestamp, 0))"""
+    )
+
+    await conn.commit()
+    logger.info("messages table rebuilt without UNIQUE constraint")
@@ -0,0 +1,45 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Enable WAL journal mode and incremental auto-vacuum.
+
+    WAL (Write-Ahead Logging):
+    - Faster writes: appends to a WAL file instead of rewriting the main DB
+    - Concurrent reads during writes (readers don't block writers)
+    - No journal file create/delete churn on every commit
+
+    Incremental auto-vacuum:
+    - Pages freed by DELETE become reclaimable without a full VACUUM
+    - Call PRAGMA incremental_vacuum to reclaim on demand
+    - Less overhead than FULL auto-vacuum (which reorganizes on every commit)
+
+    auto_vacuum mode change requires a VACUUM to restructure the file.
+    The VACUUM is performed before switching to WAL so it runs under the
+    current journal mode; WAL is then set as the final step.
+    """
+    # Check current auto_vacuum mode
+    cursor = await conn.execute("PRAGMA auto_vacuum")
+    row = await cursor.fetchone()
+    current_auto_vacuum = row[0] if row else 0
+
+    if current_auto_vacuum != 2:  # 2 = INCREMENTAL
+        logger.info("Switching auto_vacuum to INCREMENTAL (requires VACUUM)...")
+        await conn.execute("PRAGMA auto_vacuum = INCREMENTAL")
+        await conn.execute("VACUUM")
+        logger.info("VACUUM complete, auto_vacuum set to INCREMENTAL")
+    else:
+        logger.debug("auto_vacuum already INCREMENTAL, skipping VACUUM")
+
+    # Enable WAL mode (idempotent — returns current mode)
+    cursor = await conn.execute("PRAGMA journal_mode = WAL")
+    row = await cursor.fetchone()
+    mode = row[0] if row else "unknown"
+    logger.info("Journal mode set to %s", mode)
+
+    await conn.commit()
@@ -0,0 +1,29 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Enforce minimum 1-hour advert interval.
+
+    Any advert_interval between 1 and 3599 is clamped up to 3600 (1 hour).
+    Zero (disabled) is left unchanged.
+    """
+    # Guard: app_settings table may not exist if running against a very old schema
+    # (it's created in migration 9). The UPDATE is harmless if the table exists
+    # but has no rows, but will error if the table itself is missing.
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='app_settings'"
+    )
+    if await cursor.fetchone() is None:
+        logger.debug("app_settings table does not exist yet, skipping advert_interval clamp")
+        return
+
+    await conn.execute(
+        "UPDATE app_settings SET advert_interval = 3600 WHERE advert_interval > 0 AND advert_interval < 3600"
+    )
+    await conn.commit()
+    logger.debug("Clamped advert_interval to minimum 3600 seconds")
@@ -0,0 +1,33 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Create table for recent unique advert paths per repeater.
+
+    This keeps path diversity for repeater advertisements without changing the
+    existing payload-hash raw packet dedup policy.
+    """
+    await conn.execute("""
+        CREATE TABLE IF NOT EXISTS repeater_advert_paths (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            repeater_key TEXT NOT NULL,
+            path_hex TEXT NOT NULL,
+            path_len INTEGER NOT NULL,
+            first_seen INTEGER NOT NULL,
+            last_seen INTEGER NOT NULL,
+            heard_count INTEGER NOT NULL DEFAULT 1,
+            UNIQUE(repeater_key, path_hex),
+            FOREIGN KEY (repeater_key) REFERENCES contacts(public_key)
+        )
+    """)
+    await conn.execute(
+        "CREATE INDEX IF NOT EXISTS idx_repeater_advert_paths_recent "
+        "ON repeater_advert_paths(repeater_key, last_seen DESC)"
+    )
+    await conn.commit()
+    logger.debug("Ensured repeater_advert_paths table and indexes exist")
@@ -0,0 +1,60 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add first_seen column to contacts table.
+
+    Backfill strategy:
+    1. Set first_seen = last_seen for all contacts (baseline).
+    2. For contacts with PRIV messages, set first_seen = MIN(messages.received_at)
+       if that timestamp is earlier.
+    """
+    # Guard: skip if contacts table doesn't exist (e.g. partial test schemas)
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='contacts'"
+    )
+    if not await cursor.fetchone():
+        return
+
+    try:
+        await conn.execute("ALTER TABLE contacts ADD COLUMN first_seen INTEGER")
+        logger.debug("Added first_seen to contacts table")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column name" in str(e).lower():
+            logger.debug("contacts.first_seen already exists, skipping")
+        else:
+            raise
+
+    # Baseline: set first_seen = last_seen for all contacts
+    # Check if last_seen column exists (should in production, may not in minimal test schemas)
+    cursor = await conn.execute("PRAGMA table_info(contacts)")
+    columns = {row[1] for row in await cursor.fetchall()}
+    if "last_seen" in columns:
+        await conn.execute("UPDATE contacts SET first_seen = last_seen WHERE first_seen IS NULL")
+
+    # Refine: for contacts with PRIV messages, use earliest message timestamp if earlier
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='messages'"
+    )
+    if await cursor.fetchone():
+        await conn.execute(
+            """
+            UPDATE contacts SET first_seen = (
+                SELECT MIN(m.received_at) FROM messages m
+                WHERE m.type = 'PRIV' AND m.conversation_key = contacts.public_key
+            )
+            WHERE EXISTS (
+                SELECT 1 FROM messages m
+                WHERE m.type = 'PRIV' AND m.conversation_key = contacts.public_key
+                  AND m.received_at < COALESCE(contacts.first_seen, 9999999999)
+            )
+            """
+        )
+
+    await conn.commit()
+    logger.debug("Added and backfilled first_seen column")
@@ -0,0 +1,53 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Create contact_name_history table and seed with current contact names.
+    """
+    await conn.execute(
+        """
+        CREATE TABLE IF NOT EXISTS contact_name_history (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            public_key TEXT NOT NULL,
+            name TEXT NOT NULL,
+            first_seen INTEGER NOT NULL,
+            last_seen INTEGER NOT NULL,
+            UNIQUE(public_key, name),
+            FOREIGN KEY (public_key) REFERENCES contacts(public_key)
+        )
+        """
+    )
+    await conn.execute(
+        "CREATE INDEX IF NOT EXISTS idx_contact_name_history_key "
+        "ON contact_name_history(public_key, last_seen DESC)"
+    )
+
+    # Seed: one row per contact from current data (skip if contacts table doesn't exist
+    # or lacks needed columns)
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='contacts'"
+    )
+    if await cursor.fetchone():
+        cursor = await conn.execute("PRAGMA table_info(contacts)")
+        cols = {row[1] for row in await cursor.fetchall()}
+        if "name" in cols and "public_key" in cols:
+            first_seen_expr = "first_seen" if "first_seen" in cols else "0"
+            last_seen_expr = "last_seen" if "last_seen" in cols else "0"
+            await conn.execute(
+                f"""
+                INSERT OR IGNORE INTO contact_name_history (public_key, name, first_seen, last_seen)
+                SELECT public_key, name,
+                       COALESCE({first_seen_expr}, {last_seen_expr}, 0),
+                       COALESCE({last_seen_expr}, 0)
+                FROM contacts
+                WHERE name IS NOT NULL AND name != ''
+                """
+            )
+
+    await conn.commit()
+    logger.debug("Created contact_name_history table and seeded from contacts")
@@ -0,0 +1,124 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add sender_name and sender_key columns to messages table.
+
+    Backfill:
+    - sender_name for CHAN messages: extract from "Name: message" format
+    - sender_key for CHAN messages: match name to contact (skip ambiguous)
+    - sender_key for incoming PRIV messages: set to conversation_key
+    """
+    # Guard: skip if messages table doesn't exist
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='messages'"
+    )
+    if not await cursor.fetchone():
+        return
+
+    for column in ["sender_name", "sender_key"]:
+        try:
+            await conn.execute(f"ALTER TABLE messages ADD COLUMN {column} TEXT")
+            logger.debug("Added %s to messages table", column)
+        except aiosqlite.OperationalError as e:
+            if "duplicate column name" in str(e).lower():
+                logger.debug("messages.%s already exists, skipping", column)
+            else:
+                raise
+
+    # Check which columns the messages table has (may be minimal in test environments)
+    cursor = await conn.execute("PRAGMA table_info(messages)")
+    msg_cols = {row[1] for row in await cursor.fetchall()}
+
+    # Only backfill if the required columns exist
+    if "type" in msg_cols and "text" in msg_cols:
+        # Count messages to backfill for progress reporting
+        cursor = await conn.execute(
+            "SELECT COUNT(*) FROM messages WHERE type = 'CHAN' AND sender_name IS NULL"
+        )
+        row = await cursor.fetchone()
+        chan_count = row[0] if row else 0
+        if chan_count > 0:
+            logger.info("Backfilling sender_name for %d channel messages...", chan_count)
+
+        # Backfill sender_name for CHAN messages from "Name: message" format
+        # Only extract if colon position is valid (> 1 and < 51, i.e. name is 1-50 chars)
+        cursor = await conn.execute(
+            """
+            UPDATE messages SET sender_name = SUBSTR(text, 1, INSTR(text, ': ') - 1)
+            WHERE type = 'CHAN' AND sender_name IS NULL
+              AND INSTR(text, ': ') > 1 AND INSTR(text, ': ') < 52
+            """
+        )
+        if cursor.rowcount > 0:
+            logger.info("Backfilled sender_name for %d channel messages", cursor.rowcount)
+
+        # Backfill sender_key for incoming PRIV messages
+        if "outgoing" in msg_cols and "conversation_key" in msg_cols:
+            cursor = await conn.execute(
+                """
+                UPDATE messages SET sender_key = conversation_key
+                WHERE type = 'PRIV' AND outgoing = 0 AND sender_key IS NULL
+                """
+            )
+            if cursor.rowcount > 0:
+                logger.info("Backfilled sender_key for %d DM messages", cursor.rowcount)
+
+        # Backfill sender_key for CHAN messages: match sender_name to contacts
+        # Build name->key map, skip ambiguous names (multiple contacts with same name)
+        cursor = await conn.execute(
+            "SELECT name FROM sqlite_master WHERE type='table' AND name='contacts'"
+        )
+        if await cursor.fetchone():
+            cursor = await conn.execute(
+                "SELECT public_key, name FROM contacts WHERE name IS NOT NULL AND name != ''"
+            )
+            rows = await cursor.fetchall()
+
+            name_to_keys: dict[str, list[str]] = {}
+            for row in rows:
+                name = row["name"]
+                key = row["public_key"]
+                if name not in name_to_keys:
+                    name_to_keys[name] = []
+                name_to_keys[name].append(key)
+
+            # Only use unambiguous names (single contact per name)
+            unambiguous = {n: ks[0] for n, ks in name_to_keys.items() if len(ks) == 1}
+            if unambiguous:
+                logger.info(
+                    "Matching sender_key for %d unique contact names...",
+                    len(unambiguous),
+                )
+                # Use a temp table for a single bulk UPDATE instead of N individual queries
+                await conn.execute(
+                    "CREATE TEMP TABLE _name_key_map (name TEXT PRIMARY KEY, public_key TEXT NOT NULL)"
+                )
+                await conn.executemany(
+                    "INSERT INTO _name_key_map (name, public_key) VALUES (?, ?)",
+                    list(unambiguous.items()),
+                )
+                cursor = await conn.execute(
+                    """
+                    UPDATE messages SET sender_key = (
+                        SELECT public_key FROM _name_key_map WHERE _name_key_map.name = messages.sender_name
+                    )
+                    WHERE type = 'CHAN' AND sender_key IS NULL
+                      AND sender_name IN (SELECT name FROM _name_key_map)
+                    """
+                )
+                updated = cursor.rowcount
+                await conn.execute("DROP TABLE _name_key_map")
+                if updated > 0:
+                    logger.info("Backfilled sender_key for %d channel messages", updated)
+
+    # Create index on sender_key for per-contact channel message counts
+    await conn.execute("CREATE INDEX IF NOT EXISTS idx_messages_sender_key ON messages(sender_key)")
+
+    await conn.commit()
+    logger.debug("Added sender_name and sender_key columns with backfill")
@@ -0,0 +1,81 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Rename repeater_advert_paths to contact_advert_paths with column
+    repeater_key -> public_key.
+
+    Uses table rebuild since ALTER TABLE RENAME COLUMN may not be available
+    in older SQLite versions.
+    """
+    # Check if old table exists
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='repeater_advert_paths'"
+    )
+    if not await cursor.fetchone():
+        # Already renamed or doesn't exist — ensure new table exists
+        await conn.execute(
+            """
+            CREATE TABLE IF NOT EXISTS contact_advert_paths (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                public_key TEXT NOT NULL,
+                path_hex TEXT NOT NULL,
+                path_len INTEGER NOT NULL,
+                first_seen INTEGER NOT NULL,
+                last_seen INTEGER NOT NULL,
+                heard_count INTEGER NOT NULL DEFAULT 1,
+                UNIQUE(public_key, path_hex, path_len),
+                FOREIGN KEY (public_key) REFERENCES contacts(public_key)
+            )
+            """
+        )
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_contact_advert_paths_recent "
+            "ON contact_advert_paths(public_key, last_seen DESC)"
+        )
+        await conn.commit()
+        logger.debug("contact_advert_paths already exists or old table missing, skipping rename")
+        return
+
+    # Create new table (IF NOT EXISTS in case SCHEMA already created it)
+    await conn.execute(
+        """
+        CREATE TABLE IF NOT EXISTS contact_advert_paths (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            public_key TEXT NOT NULL,
+            path_hex TEXT NOT NULL,
+            path_len INTEGER NOT NULL,
+            first_seen INTEGER NOT NULL,
+            last_seen INTEGER NOT NULL,
+            heard_count INTEGER NOT NULL DEFAULT 1,
+            UNIQUE(public_key, path_hex, path_len),
+            FOREIGN KEY (public_key) REFERENCES contacts(public_key)
+        )
+        """
+    )
+
+    # Copy data (INSERT OR IGNORE in case of duplicates)
+    await conn.execute(
+        """
+        INSERT OR IGNORE INTO contact_advert_paths (public_key, path_hex, path_len, first_seen, last_seen, heard_count)
+        SELECT repeater_key, path_hex, path_len, first_seen, last_seen, heard_count
+        FROM repeater_advert_paths
+        """
+    )
+
+    # Drop old table
+    await conn.execute("DROP TABLE repeater_advert_paths")
+
+    # Create index
+    await conn.execute(
+        "CREATE INDEX IF NOT EXISTS idx_contact_advert_paths_recent "
+        "ON contact_advert_paths(public_key, last_seen DESC)"
+    )
+
+    await conn.commit()
+    logger.info("Renamed repeater_advert_paths to contact_advert_paths")
@@ -0,0 +1,36 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Backfill contacts.first_seen from contact_advert_paths where advert path
+    first_seen is earlier than the contact's current first_seen.
+    """
+    # Guard: skip if either table doesn't exist
+    for table in ("contacts", "contact_advert_paths"):
+        cursor = await conn.execute(
+            "SELECT name FROM sqlite_master WHERE type='table' AND name=?", (table,)
+        )
+        if not await cursor.fetchone():
+            return
+
+    await conn.execute(
+        """
+        UPDATE contacts SET first_seen = (
+            SELECT MIN(cap.first_seen) FROM contact_advert_paths cap
+            WHERE cap.public_key = contacts.public_key
+        )
+        WHERE EXISTS (
+            SELECT 1 FROM contact_advert_paths cap
+            WHERE cap.public_key = contacts.public_key
+              AND cap.first_seen < COALESCE(contacts.first_seen, 9999999999)
+        )
+        """
+    )
+
+    await conn.commit()
+    logger.debug("Backfilled first_seen from contact_advert_paths")
@@ -0,0 +1,107 @@
+import logging
+from hashlib import sha256
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Convert payload_hash from 64-char hex TEXT to 32-byte BLOB.
+
+    Halves storage for both the column data and its UNIQUE index.
+    Uses Python bytes.fromhex() for the conversion since SQLite's unhex()
+    requires 3.41.0+ which may not be available on all deployments.
+    """
+    # Guard: skip if raw_packets table doesn't exist
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='raw_packets'"
+    )
+    if not await cursor.fetchone():
+        logger.debug("raw_packets table does not exist, skipping payload_hash conversion")
+        await conn.commit()
+        return
+
+    # Check column types — skip if payload_hash doesn't exist or is already BLOB
+    cursor = await conn.execute("PRAGMA table_info(raw_packets)")
+    cols = {row[1]: row[2] for row in await cursor.fetchall()}
+    if "payload_hash" not in cols:
+        logger.debug("payload_hash column does not exist, skipping conversion")
+        await conn.commit()
+        return
+    if cols["payload_hash"].upper() == "BLOB":
+        logger.debug("payload_hash is already BLOB, skipping conversion")
+        await conn.commit()
+        return
+
+    logger.info("Rebuilding raw_packets to convert payload_hash TEXT → BLOB...")
+
+    # Create new table with BLOB type
+    await conn.execute("""
+        CREATE TABLE raw_packets_new (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            timestamp INTEGER NOT NULL,
+            data BLOB NOT NULL,
+            message_id INTEGER,
+            payload_hash BLOB,
+            FOREIGN KEY (message_id) REFERENCES messages(id)
+        )
+    """)
+
+    # Batch-convert rows: read TEXT hashes, convert to bytes, insert into new table
+    batch_size = 5000
+    cursor = await conn.execute(
+        "SELECT id, timestamp, data, message_id, payload_hash FROM raw_packets ORDER BY id"
+    )
+
+    total = 0
+    while True:
+        rows = await cursor.fetchmany(batch_size)
+        if not rows:
+            break
+
+        batch: list[tuple[int, int, bytes, int | None, bytes | None]] = []
+        for row in rows:
+            rid, ts, data, mid, ph = row[0], row[1], row[2], row[3], row[4]
+            if ph is not None and isinstance(ph, str):
+                try:
+                    ph = bytes.fromhex(ph)
+                except ValueError:
+                    # Not a valid hex string — hash the value to produce a valid BLOB
+                    ph = sha256(ph.encode()).digest()
+            batch.append((rid, ts, data, mid, ph))
+
+        await conn.executemany(
+            "INSERT INTO raw_packets_new (id, timestamp, data, message_id, payload_hash) "
+            "VALUES (?, ?, ?, ?, ?)",
+            batch,
+        )
+        total += len(batch)
+
+        if total % 50000 == 0:
+            logger.info("Converted %d rows...", total)
+
+    # Preserve autoincrement sequence
+    cursor = await conn.execute("SELECT seq FROM sqlite_sequence WHERE name = 'raw_packets'")
+    seq_row = await cursor.fetchone()
+    if seq_row is not None:
+        await conn.execute(
+            "INSERT OR REPLACE INTO sqlite_sequence (name, seq) VALUES ('raw_packets_new', ?)",
+            (seq_row[0],),
+        )
+
+    await conn.execute("DROP TABLE raw_packets")
+    await conn.execute("ALTER TABLE raw_packets_new RENAME TO raw_packets")
+
+    # Clean up the sqlite_sequence entry for the old temp name
+    await conn.execute("DELETE FROM sqlite_sequence WHERE name = 'raw_packets_new'")
+
+    # Recreate indexes
+    await conn.execute(
+        "CREATE UNIQUE INDEX idx_raw_packets_payload_hash ON raw_packets(payload_hash)"
+    )
+    await conn.execute("CREATE INDEX idx_raw_packets_message_id ON raw_packets(message_id)")
+
+    await conn.commit()
+    logger.info("Converted %d payload_hash values from TEXT to BLOB", total)
@@ -0,0 +1,27 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add a covering index for the unread counts query.
+
+    The /api/read-state/unreads endpoint runs three queries against messages.
+    The last-message-times query (GROUP BY type, conversation_key + MAX(received_at))
+    was doing a full table scan. This covering index lets SQLite resolve the
+    grouping and MAX entirely from the index without touching the table.
+    It also improves the unread count queries which filter on outgoing and received_at.
+    """
+    # Guard: table or columns may not exist in partial-schema test setups
+    cursor = await conn.execute("PRAGMA table_info(messages)")
+    columns = {row[1] for row in await cursor.fetchall()}
+    required = {"type", "conversation_key", "outgoing", "received_at"}
+    if required <= columns:
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_messages_unread_covering "
+            "ON messages(type, conversation_key, outgoing, received_at)"
+        )
+    await conn.commit()
@@ -0,0 +1,31 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """
+    Add a composite index for message pagination and drop the now-redundant
+    idx_messages_conversation.
+
+    The pagination query (ORDER BY received_at DESC, id DESC LIMIT N) hits a
+    temp B-tree sort without this index. With it, SQLite walks the index in
+    order and stops after N rows — critical for channels with 30K+ messages.
+
+    idx_messages_conversation(type, conversation_key) is a strict prefix of
+    both this index and idx_messages_unread_covering, so SQLite never picks it.
+    Dropping it saves ~6 MB and one index to maintain per INSERT.
+    """
+    # Guard: table or columns may not exist in partial-schema test setups
+    cursor = await conn.execute("PRAGMA table_info(messages)")
+    columns = {row[1] for row in await cursor.fetchall()}
+    required = {"type", "conversation_key", "received_at", "id"}
+    if required <= columns:
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_messages_pagination "
+            "ON messages(type, conversation_key, received_at DESC, id DESC)"
+        )
+        await conn.execute("DROP INDEX IF EXISTS idx_messages_conversation")
+    await conn.commit()
@@ -0,0 +1,37 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add MQTT configuration columns to app_settings."""
+    # Guard: app_settings may not exist in partial-schema test setups
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='app_settings'"
+    )
+    if not await cursor.fetchone():
+        await conn.commit()
+        return
+
+    cursor = await conn.execute("PRAGMA table_info(app_settings)")
+    columns = {row[1] for row in await cursor.fetchall()}
+
+    new_columns = [
+        ("mqtt_broker_host", "TEXT DEFAULT ''"),
+        ("mqtt_broker_port", "INTEGER DEFAULT 1883"),
+        ("mqtt_username", "TEXT DEFAULT ''"),
+        ("mqtt_password", "TEXT DEFAULT ''"),
+        ("mqtt_use_tls", "INTEGER DEFAULT 0"),
+        ("mqtt_tls_insecure", "INTEGER DEFAULT 0"),
+        ("mqtt_topic_prefix", "TEXT DEFAULT 'meshcore'"),
+        ("mqtt_publish_messages", "INTEGER DEFAULT 0"),
+        ("mqtt_publish_raw_packets", "INTEGER DEFAULT 0"),
+    ]
+
+    for col_name, col_def in new_columns:
+        if col_name not in columns:
+            await conn.execute(f"ALTER TABLE app_settings ADD COLUMN {col_name} {col_def}")
+
+    await conn.commit()
@@ -0,0 +1,33 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add community MQTT configuration columns to app_settings."""
+    # Guard: app_settings may not exist in partial-schema test setups
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='app_settings'"
+    )
+    if not await cursor.fetchone():
+        await conn.commit()
+        return
+
+    cursor = await conn.execute("PRAGMA table_info(app_settings)")
+    columns = {row[1] for row in await cursor.fetchall()}
+
+    new_columns = [
+        ("community_mqtt_enabled", "INTEGER DEFAULT 0"),
+        ("community_mqtt_iata", "TEXT DEFAULT ''"),
+        ("community_mqtt_broker_host", "TEXT DEFAULT 'mqtt-us-v1.letsmesh.net'"),
+        ("community_mqtt_broker_port", "INTEGER DEFAULT 443"),
+        ("community_mqtt_email", "TEXT DEFAULT ''"),
+    ]
+
+    for col_name, col_def in new_columns:
+        if col_name not in columns:
+            await conn.execute(f"ALTER TABLE app_settings ADD COLUMN {col_name} {col_def}")
+
+    await conn.commit()
@@ -0,0 +1,23 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Seed the #remoteterm hashtag channel so new installs have it by default.
+
+    Uses INSERT OR IGNORE so it's a no-op if the channel already exists
+    (e.g. existing users who already added it manually). The channels table
+    is created by the base schema before migrations run, so it always exists
+    in production.
+    """
+    try:
+        await conn.execute(
+            "INSERT OR IGNORE INTO channels (key, name, is_hashtag, on_radio) VALUES (?, ?, ?, ?)",
+            ("8959AE053F2201801342A1DBDDA184F6", "#remoteterm", 1, 0),
+        )
+        await conn.commit()
+    except Exception:
+        logger.debug("Skipping #remoteterm seed (channels table not ready)")
@@ -0,0 +1,23 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add flood_scope column to app_settings for outbound region tagging.
+
+    Empty string means disabled (no scope set, messages sent unscoped).
+    """
+    try:
+        await conn.execute("ALTER TABLE app_settings ADD COLUMN flood_scope TEXT DEFAULT ''")
+        await conn.commit()
+    except Exception as e:
+        error_msg = str(e).lower()
+        if "duplicate column" in error_msg:
+            logger.debug("flood_scope column already exists, skipping")
+        elif "no such table" in error_msg:
+            logger.debug("app_settings table not ready, skipping flood_scope migration")
+        else:
+            raise
@@ -0,0 +1,36 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add blocked_keys and blocked_names columns to app_settings.
+
+    These store JSON arrays of blocked public keys and display names.
+    Blocking hides messages from the UI but does not affect MQTT or bots.
+    """
+    try:
+        await conn.execute("ALTER TABLE app_settings ADD COLUMN blocked_keys TEXT DEFAULT '[]'")
+    except Exception as e:
+        error_msg = str(e).lower()
+        if "duplicate column" in error_msg:
+            logger.debug("blocked_keys column already exists, skipping")
+        elif "no such table" in error_msg:
+            logger.debug("app_settings table not ready, skipping blocked_keys migration")
+        else:
+            raise
+
+    try:
+        await conn.execute("ALTER TABLE app_settings ADD COLUMN blocked_names TEXT DEFAULT '[]'")
+    except Exception as e:
+        error_msg = str(e).lower()
+        if "duplicate column" in error_msg:
+            logger.debug("blocked_names column already exists, skipping")
+        elif "no such table" in error_msg:
+            logger.debug("app_settings table not ready, skipping blocked_names migration")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,143 @@
+import json
+import logging
+import uuid
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Create fanout_configs table and migrate existing MQTT settings.
+
+    Reads existing MQTT settings from app_settings and creates corresponding
+    fanout_configs rows. Old columns are NOT dropped (rollback safety).
+    """
+
+    # 1. Create fanout_configs table
+    await conn.execute(
+        """
+        CREATE TABLE IF NOT EXISTS fanout_configs (
+            id TEXT PRIMARY KEY,
+            type TEXT NOT NULL,
+            name TEXT NOT NULL,
+            enabled INTEGER DEFAULT 0,
+            config TEXT NOT NULL DEFAULT '{}',
+            scope TEXT NOT NULL DEFAULT '{}',
+            sort_order INTEGER DEFAULT 0,
+            created_at INTEGER NOT NULL
+        )
+        """
+    )
+
+    # 2. Read existing MQTT settings
+    try:
+        cursor = await conn.execute(
+            """
+            SELECT mqtt_broker_host, mqtt_broker_port, mqtt_username, mqtt_password,
+                   mqtt_use_tls, mqtt_tls_insecure, mqtt_topic_prefix,
+                   mqtt_publish_messages, mqtt_publish_raw_packets,
+                   community_mqtt_enabled, community_mqtt_iata,
+                   community_mqtt_broker_host, community_mqtt_broker_port,
+                   community_mqtt_email
+            FROM app_settings WHERE id = 1
+            """
+        )
+        row = await cursor.fetchone()
+    except Exception:
+        row = None
+
+    if row is None:
+        await conn.commit()
+        return
+
+    import time
+
+    now = int(time.time())
+    sort_order = 0
+
+    # 3. Migrate private MQTT if configured
+    broker_host = row["mqtt_broker_host"] or ""
+    if broker_host:
+        publish_messages = bool(row["mqtt_publish_messages"])
+        publish_raw = bool(row["mqtt_publish_raw_packets"])
+        enabled = publish_messages or publish_raw
+
+        config = {
+            "broker_host": broker_host,
+            "broker_port": row["mqtt_broker_port"] or 1883,
+            "username": row["mqtt_username"] or "",
+            "password": row["mqtt_password"] or "",
+            "use_tls": bool(row["mqtt_use_tls"]),
+            "tls_insecure": bool(row["mqtt_tls_insecure"]),
+            "topic_prefix": row["mqtt_topic_prefix"] or "meshcore",
+        }
+
+        scope = {
+            "messages": "all" if publish_messages else "none",
+            "raw_packets": "all" if publish_raw else "none",
+        }
+
+        await conn.execute(
+            """
+            INSERT INTO fanout_configs (id, type, name, enabled, config, scope, sort_order, created_at)
+            VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+            """,
+            (
+                str(uuid.uuid4()),
+                "mqtt_private",
+                "Private MQTT",
+                1 if enabled else 0,
+                json.dumps(config),
+                json.dumps(scope),
+                sort_order,
+                now,
+            ),
+        )
+        sort_order += 1
+        logger.info("Migrated private MQTT settings to fanout_configs (enabled=%s)", enabled)
+
+    # 4. Migrate community MQTT if enabled OR configured (preserve disabled-but-configured)
+    community_enabled = bool(row["community_mqtt_enabled"])
+    community_iata = row["community_mqtt_iata"] or ""
+    community_host = row["community_mqtt_broker_host"] or ""
+    community_email = row["community_mqtt_email"] or ""
+    community_has_config = bool(
+        community_iata
+        or community_email
+        or (community_host and community_host != "mqtt-us-v1.letsmesh.net")
+    )
+    if community_enabled or community_has_config:
+        config = {
+            "broker_host": community_host or "mqtt-us-v1.letsmesh.net",
+            "broker_port": row["community_mqtt_broker_port"] or 443,
+            "iata": community_iata,
+            "email": community_email,
+        }
+
+        scope = {
+            "messages": "none",
+            "raw_packets": "all",
+        }
+
+        await conn.execute(
+            """
+            INSERT INTO fanout_configs (id, type, name, enabled, config, scope, sort_order, created_at)
+            VALUES (?, ?, ?, ?, ?, ?, ?, ?)
+            """,
+            (
+                str(uuid.uuid4()),
+                "mqtt_community",
+                "Community MQTT",
+                1 if community_enabled else 0,
+                json.dumps(config),
+                json.dumps(scope),
+                sort_order,
+                now,
+            ),
+        )
+        logger.info(
+            "Migrated community MQTT settings to fanout_configs (enabled=%s)", community_enabled
+        )
+
+    await conn.commit()
@@ -0,0 +1,63 @@
+import json
+import logging
+import uuid
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Migrate bots from app_settings.bots JSON to fanout_configs rows."""
+
+    try:
+        cursor = await conn.execute("SELECT bots FROM app_settings WHERE id = 1")
+        row = await cursor.fetchone()
+    except Exception:
+        row = None
+
+    if row is None:
+        await conn.commit()
+        return
+
+    bots_json = row["bots"] or "[]"
+    try:
+        bots = json.loads(bots_json)
+    except (json.JSONDecodeError, TypeError):
+        bots = []
+
+    if not bots:
+        await conn.commit()
+        return
+
+    import time
+
+    now = int(time.time())
+
+    # Use sort_order starting at 200 to place bots after MQTT configs (0-99)
+    for i, bot in enumerate(bots):
+        bot_name = bot.get("name") or f"Bot {i + 1}"
+        bot_enabled = bool(bot.get("enabled", False))
+        bot_code = bot.get("code", "")
+
+        config_blob = json.dumps({"code": bot_code})
+        scope = json.dumps({"messages": "all", "raw_packets": "none"})
+
+        await conn.execute(
+            """
+            INSERT INTO fanout_configs (id, type, name, enabled, config, scope, sort_order, created_at)
+            VALUES (?, 'bot', ?, ?, ?, ?, ?, ?)
+            """,
+            (
+                str(uuid.uuid4()),
+                bot_name,
+                1 if bot_enabled else 0,
+                config_blob,
+                scope,
+                200 + i,
+                now,
+            ),
+        )
+        logger.info("Migrated bot '%s' to fanout_configs (enabled=%s)", bot_name, bot_enabled)
+
+    await conn.commit()
@@ -0,0 +1,54 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Drop legacy MQTT, community MQTT, and bots columns from app_settings.
+
+    These columns were migrated to fanout_configs in migrations 36 and 37.
+    SQLite 3.35.0+ supports ALTER TABLE DROP COLUMN. For older versions,
+    the columns remain but are harmless (no longer read or written).
+    """
+    # Check if app_settings table exists (some test DBs may not have it)
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='app_settings'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.commit()
+        return
+
+    columns_to_drop = [
+        "bots",
+        "mqtt_broker_host",
+        "mqtt_broker_port",
+        "mqtt_username",
+        "mqtt_password",
+        "mqtt_use_tls",
+        "mqtt_tls_insecure",
+        "mqtt_topic_prefix",
+        "mqtt_publish_messages",
+        "mqtt_publish_raw_packets",
+        "community_mqtt_enabled",
+        "community_mqtt_iata",
+        "community_mqtt_broker_host",
+        "community_mqtt_broker_port",
+        "community_mqtt_email",
+    ]
+
+    for column in columns_to_drop:
+        try:
+            await conn.execute(f"ALTER TABLE app_settings DROP COLUMN {column}")
+            logger.debug("Dropped %s from app_settings", column)
+        except aiosqlite.OperationalError as e:
+            error_msg = str(e).lower()
+            if "no such column" in error_msg:
+                logger.debug("app_settings.%s already dropped, skipping", column)
+            elif "syntax error" in error_msg or "drop column" in error_msg:
+                logger.debug("SQLite doesn't support DROP COLUMN, %s column will remain", column)
+            else:
+                raise
+
+    await conn.commit()
@@ -0,0 +1,65 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add contacts.out_path_hash_mode and backfill legacy rows.
+
+    Historical databases predate multibyte routing support. Backfill rules:
+    - contacts with last_path_len = -1 are flood routes -> out_path_hash_mode = -1
+    - all other existing contacts default to 0 (1-byte legacy hop identifiers)
+    """
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='contacts'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.commit()
+        return
+
+    column_cursor = await conn.execute("PRAGMA table_info(contacts)")
+    columns = {row[1] for row in await column_cursor.fetchall()}
+
+    added_column = False
+
+    try:
+        await conn.execute(
+            "ALTER TABLE contacts ADD COLUMN out_path_hash_mode INTEGER NOT NULL DEFAULT 0"
+        )
+        added_column = True
+        logger.debug("Added out_path_hash_mode to contacts table")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column name" in str(e).lower():
+            logger.debug("contacts.out_path_hash_mode already exists, skipping add")
+        else:
+            raise
+
+    if "last_path_len" not in columns:
+        await conn.commit()
+        return
+
+    if added_column:
+        await conn.execute(
+            """
+            UPDATE contacts
+            SET out_path_hash_mode = CASE
+                WHEN last_path_len = -1 THEN -1
+                ELSE 0
+            END
+            """
+        )
+    else:
+        await conn.execute(
+            """
+            UPDATE contacts
+            SET out_path_hash_mode = CASE
+                WHEN last_path_len = -1 THEN -1
+                ELSE 0
+            END
+            WHERE out_path_hash_mode NOT IN (-1, 0, 1, 2)
+               OR (last_path_len = -1 AND out_path_hash_mode != -1)
+            """
+        )
+    await conn.commit()
@@ -0,0 +1,82 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(
+    conn: aiosqlite.Connection,
+) -> None:
+    """Rebuild contact_advert_paths so uniqueness includes path_len.
+
+    Multi-byte routing can produce the same path_hex bytes with a different hop count,
+    which changes the hop boundaries and therefore the semantic next-hop identity.
+    """
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='contact_advert_paths'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.execute(
+            """
+            CREATE TABLE IF NOT EXISTS contact_advert_paths (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                public_key TEXT NOT NULL,
+                path_hex TEXT NOT NULL,
+                path_len INTEGER NOT NULL,
+                first_seen INTEGER NOT NULL,
+                last_seen INTEGER NOT NULL,
+                heard_count INTEGER NOT NULL DEFAULT 1,
+                UNIQUE(public_key, path_hex, path_len),
+                FOREIGN KEY (public_key) REFERENCES contacts(public_key)
+            )
+            """
+        )
+        await conn.execute("DROP INDEX IF EXISTS idx_contact_advert_paths_recent")
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_contact_advert_paths_recent "
+            "ON contact_advert_paths(public_key, last_seen DESC)"
+        )
+        await conn.commit()
+        return
+
+    await conn.execute(
+        """
+        CREATE TABLE contact_advert_paths_new (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            public_key TEXT NOT NULL,
+            path_hex TEXT NOT NULL,
+            path_len INTEGER NOT NULL,
+            first_seen INTEGER NOT NULL,
+            last_seen INTEGER NOT NULL,
+            heard_count INTEGER NOT NULL DEFAULT 1,
+            UNIQUE(public_key, path_hex, path_len),
+            FOREIGN KEY (public_key) REFERENCES contacts(public_key)
+        )
+        """
+    )
+
+    await conn.execute(
+        """
+        INSERT INTO contact_advert_paths_new
+            (public_key, path_hex, path_len, first_seen, last_seen, heard_count)
+        SELECT
+            public_key,
+            path_hex,
+            path_len,
+            MIN(first_seen),
+            MAX(last_seen),
+            SUM(heard_count)
+        FROM contact_advert_paths
+        GROUP BY public_key, path_hex, path_len
+        """
+    )
+
+    await conn.execute("DROP TABLE contact_advert_paths")
+    await conn.execute("ALTER TABLE contact_advert_paths_new RENAME TO contact_advert_paths")
+    await conn.execute("DROP INDEX IF EXISTS idx_contact_advert_paths_recent")
+    await conn.execute(
+        "CREATE INDEX IF NOT EXISTS idx_contact_advert_paths_recent "
+        "ON contact_advert_paths(public_key, last_seen DESC)"
+    )
+    await conn.commit()
@@ -0,0 +1,31 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add nullable routing-override columns to contacts."""
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='contacts'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.commit()
+        return
+
+    for column_name, column_type in (
+        ("route_override_path", "TEXT"),
+        ("route_override_len", "INTEGER"),
+        ("route_override_hash_mode", "INTEGER"),
+    ):
+        try:
+            await conn.execute(f"ALTER TABLE contacts ADD COLUMN {column_name} {column_type}")
+            logger.debug("Added %s to contacts table", column_name)
+        except aiosqlite.OperationalError as e:
+            if "duplicate column name" in str(e).lower():
+                logger.debug("contacts.%s already exists, skipping", column_name)
+            else:
+                raise
+
+    await conn.commit()
@@ -0,0 +1,26 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add nullable per-channel flood-scope override column."""
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='channels'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.commit()
+        return
+
+    try:
+        await conn.execute("ALTER TABLE channels ADD COLUMN flood_scope_override TEXT")
+        logger.debug("Added flood_scope_override to channels table")
+    except aiosqlite.OperationalError as e:
+        if "duplicate column name" in str(e).lower():
+            logger.debug("channels.flood_scope_override already exists, skipping")
+        else:
+            raise
+
+    await conn.commit()
@@ -0,0 +1,31 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Restrict the message dedup index to channel messages."""
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='messages'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.commit()
+        return
+
+    cursor = await conn.execute("PRAGMA table_info(messages)")
+    columns = {row[1] for row in await cursor.fetchall()}
+    required_columns = {"type", "conversation_key", "text", "sender_timestamp"}
+    if not required_columns.issubset(columns):
+        logger.debug("messages table missing dedup-index columns, skipping migration 43")
+        await conn.commit()
+        return
+
+    await conn.execute("DROP INDEX IF EXISTS idx_messages_dedup_null_safe")
+    await conn.execute(
+        """CREATE UNIQUE INDEX IF NOT EXISTS idx_messages_dedup_null_safe
+           ON messages(type, conversation_key, text, COALESCE(sender_timestamp, 0))
+           WHERE type = 'CHAN'"""
+    )
+    await conn.commit()
@@ -0,0 +1,157 @@
+import json
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+def _merge_message_paths(paths_json_values: list[str | None]) -> str | None:
+    """Merge multiple message path arrays into one exact-observation list."""
+    merged: list[dict[str, object]] = []
+    seen: set[tuple[object | None, object | None, object | None]] = set()
+
+    for paths_json in paths_json_values:
+        if not paths_json:
+            continue
+        try:
+            parsed = json.loads(paths_json)
+        except (TypeError, json.JSONDecodeError):
+            continue
+        if not isinstance(parsed, list):
+            continue
+        for entry in parsed:
+            if not isinstance(entry, dict):
+                continue
+            key = (
+                entry.get("path"),
+                entry.get("received_at"),
+                entry.get("path_len"),
+            )
+            if key in seen:
+                continue
+            seen.add(key)
+            merged.append(entry)
+
+    return json.dumps(merged) if merged else None
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Collapse same-contact same-text same-second incoming DMs into one row."""
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='messages'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.commit()
+        return
+
+    cursor = await conn.execute("PRAGMA table_info(messages)")
+    columns = {row[1] for row in await cursor.fetchall()}
+    required_columns = {
+        "id",
+        "type",
+        "conversation_key",
+        "text",
+        "sender_timestamp",
+        "received_at",
+        "paths",
+        "txt_type",
+        "signature",
+        "outgoing",
+        "acked",
+        "sender_name",
+        "sender_key",
+    }
+    if not required_columns.issubset(columns):
+        logger.debug("messages table missing incoming-DM dedup columns, skipping migration 44")
+        await conn.commit()
+        return
+
+    raw_packets_cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='raw_packets'"
+    )
+    raw_packets_exists = await raw_packets_cursor.fetchone() is not None
+
+    duplicate_groups_cursor = await conn.execute(
+        """
+        SELECT conversation_key, text,
+               COALESCE(sender_timestamp, 0) AS normalized_sender_timestamp,
+               COUNT(*) AS duplicate_count
+        FROM messages
+        WHERE type = 'PRIV' AND outgoing = 0
+        GROUP BY conversation_key, text, COALESCE(sender_timestamp, 0)
+        HAVING COUNT(*) > 1
+        """
+    )
+    duplicate_groups = await duplicate_groups_cursor.fetchall()
+
+    for group in duplicate_groups:
+        normalized_sender_timestamp = group["normalized_sender_timestamp"]
+        rows_cursor = await conn.execute(
+            """
+            SELECT *
+            FROM messages
+            WHERE type = 'PRIV' AND outgoing = 0
+              AND conversation_key = ? AND text = ?
+              AND COALESCE(sender_timestamp, 0) = ?
+            ORDER BY id ASC
+            """,
+            (
+                group["conversation_key"],
+                group["text"],
+                normalized_sender_timestamp,
+            ),
+        )
+        rows = list(await rows_cursor.fetchall())
+        if len(rows) < 2:
+            continue
+
+        keeper = rows[0]
+        duplicate_ids = [row["id"] for row in rows[1:]]
+        merged_paths = _merge_message_paths([row["paths"] for row in rows])
+        merged_received_at = min(row["received_at"] for row in rows)
+        merged_txt_type = next((row["txt_type"] for row in rows if row["txt_type"] != 0), 0)
+        merged_signature = next((row["signature"] for row in rows if row["signature"]), None)
+        merged_sender_name = next((row["sender_name"] for row in rows if row["sender_name"]), None)
+        merged_sender_key = next((row["sender_key"] for row in rows if row["sender_key"]), None)
+        merged_acked = max(int(row["acked"] or 0) for row in rows)
+
+        await conn.execute(
+            """
+            UPDATE messages
+            SET received_at = ?, paths = ?, txt_type = ?, signature = ?,
+                acked = ?, sender_name = ?, sender_key = ?
+            WHERE id = ?
+            """,
+            (
+                merged_received_at,
+                merged_paths,
+                merged_txt_type,
+                merged_signature,
+                merged_acked,
+                merged_sender_name,
+                merged_sender_key,
+                keeper["id"],
+            ),
+        )
+
+        if raw_packets_exists:
+            for duplicate_id in duplicate_ids:
+                await conn.execute(
+                    "UPDATE raw_packets SET message_id = ? WHERE message_id = ?",
+                    (keeper["id"], duplicate_id),
+                )
+
+        placeholders = ",".join("?" for _ in duplicate_ids)
+        await conn.execute(
+            f"DELETE FROM messages WHERE id IN ({placeholders})",
+            duplicate_ids,
+        )
+
+    await conn.execute("DROP INDEX IF EXISTS idx_messages_incoming_priv_dedup")
+    await conn.execute(
+        """CREATE UNIQUE INDEX IF NOT EXISTS idx_messages_incoming_priv_dedup
+           ON messages(type, conversation_key, text, COALESCE(sender_timestamp, 0))
+           WHERE type = 'PRIV' AND outgoing = 0"""
+    )
+    await conn.commit()
@@ -0,0 +1,136 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Replace legacy contact route columns with canonical direct-route columns."""
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='contacts'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.commit()
+        return
+
+    cursor = await conn.execute("PRAGMA table_info(contacts)")
+    columns = {row[1] for row in await cursor.fetchall()}
+
+    target_columns = {
+        "public_key",
+        "name",
+        "type",
+        "flags",
+        "direct_path",
+        "direct_path_len",
+        "direct_path_hash_mode",
+        "direct_path_updated_at",
+        "route_override_path",
+        "route_override_len",
+        "route_override_hash_mode",
+        "last_advert",
+        "lat",
+        "lon",
+        "last_seen",
+        "on_radio",
+        "last_contacted",
+        "first_seen",
+        "last_read_at",
+    }
+    if (
+        target_columns.issubset(columns)
+        and "last_path" not in columns
+        and "out_path_hash_mode" not in columns
+    ):
+        await conn.commit()
+        return
+
+    await conn.execute(
+        """
+        CREATE TABLE contacts_new (
+            public_key TEXT PRIMARY KEY,
+            name TEXT,
+            type INTEGER DEFAULT 0,
+            flags INTEGER DEFAULT 0,
+            direct_path TEXT,
+            direct_path_len INTEGER,
+            direct_path_hash_mode INTEGER,
+            direct_path_updated_at INTEGER,
+            route_override_path TEXT,
+            route_override_len INTEGER,
+            route_override_hash_mode INTEGER,
+            last_advert INTEGER,
+            lat REAL,
+            lon REAL,
+            last_seen INTEGER,
+            on_radio INTEGER DEFAULT 0,
+            last_contacted INTEGER,
+            first_seen INTEGER,
+            last_read_at INTEGER
+        )
+        """
+    )
+
+    select_expr = {
+        "public_key": "public_key",
+        "name": "NULL",
+        "type": "0",
+        "flags": "0",
+        "direct_path": "NULL",
+        "direct_path_len": "NULL",
+        "direct_path_hash_mode": "NULL",
+        "direct_path_updated_at": "NULL",
+        "route_override_path": "NULL",
+        "route_override_len": "NULL",
+        "route_override_hash_mode": "NULL",
+        "last_advert": "NULL",
+        "lat": "NULL",
+        "lon": "NULL",
+        "last_seen": "NULL",
+        "on_radio": "0",
+        "last_contacted": "NULL",
+        "first_seen": "NULL",
+        "last_read_at": "NULL",
+    }
+    for name in ("name", "type", "flags"):
+        if name in columns:
+            select_expr[name] = name
+
+    if "direct_path" in columns:
+        select_expr["direct_path"] = "direct_path"
+
+    if "direct_path_len" in columns:
+        select_expr["direct_path_len"] = "direct_path_len"
+
+    if "direct_path_hash_mode" in columns:
+        select_expr["direct_path_hash_mode"] = "direct_path_hash_mode"
+
+    for name in (
+        "route_override_path",
+        "route_override_len",
+        "route_override_hash_mode",
+        "last_advert",
+        "lat",
+        "lon",
+        "last_seen",
+        "on_radio",
+        "last_contacted",
+        "first_seen",
+        "last_read_at",
+    ):
+        if name in columns:
+            select_expr[name] = name
+
+    ordered_columns = list(select_expr.keys())
+    await conn.execute(
+        f"""
+        INSERT INTO contacts_new ({", ".join(ordered_columns)})
+        SELECT {", ".join(select_expr[name] for name in ordered_columns)}
+        FROM contacts
+        """
+    )
+
+    await conn.execute("DROP TABLE contacts")
+    await conn.execute("ALTER TABLE contacts_new RENAME TO contacts")
+    await conn.commit()
@@ -0,0 +1,93 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Move uniquely resolvable orphan contact child rows onto full contacts, drop the rest."""
+    existing_tables_cursor = await conn.execute("SELECT name FROM sqlite_master WHERE type='table'")
+    existing_tables = {row[0] for row in await existing_tables_cursor.fetchall()}
+    if "contacts" not in existing_tables:
+        await conn.commit()
+        return
+
+    child_tables = [
+        table
+        for table in ("contact_name_history", "contact_advert_paths")
+        if table in existing_tables
+    ]
+    if not child_tables:
+        await conn.commit()
+        return
+
+    orphan_keys: set[str] = set()
+
+    for table in child_tables:
+        cursor = await conn.execute(
+            f"""
+            SELECT DISTINCT child.public_key
+            FROM {table} child
+            LEFT JOIN contacts c ON c.public_key = child.public_key
+            WHERE c.public_key IS NULL
+            """
+        )
+        orphan_keys.update(row[0] for row in await cursor.fetchall())
+
+    for orphan_key in sorted(orphan_keys, key=len, reverse=True):
+        match_cursor = await conn.execute(
+            """
+            SELECT public_key
+            FROM contacts
+            WHERE length(public_key) = 64
+              AND public_key LIKE ? || '%'
+            ORDER BY public_key
+            """,
+            (orphan_key.lower(),),
+        )
+        matches = [row[0] for row in await match_cursor.fetchall()]
+        resolved_key = matches[0] if len(matches) == 1 else None
+
+        if resolved_key is not None:
+            if "contact_name_history" in child_tables:
+                await conn.execute(
+                    """
+                    INSERT INTO contact_name_history (public_key, name, first_seen, last_seen)
+                    SELECT ?, name, first_seen, last_seen
+                    FROM contact_name_history
+                    WHERE public_key = ?
+                    ON CONFLICT(public_key, name) DO UPDATE SET
+                        first_seen = MIN(contact_name_history.first_seen, excluded.first_seen),
+                        last_seen = MAX(contact_name_history.last_seen, excluded.last_seen)
+                    """,
+                    (resolved_key, orphan_key),
+                )
+            if "contact_advert_paths" in child_tables:
+                await conn.execute(
+                    """
+                    INSERT INTO contact_advert_paths
+                        (public_key, path_hex, path_len, first_seen, last_seen, heard_count)
+                    SELECT ?, path_hex, path_len, first_seen, last_seen, heard_count
+                    FROM contact_advert_paths
+                    WHERE public_key = ?
+                    ON CONFLICT(public_key, path_hex, path_len) DO UPDATE SET
+                        first_seen = MIN(contact_advert_paths.first_seen, excluded.first_seen),
+                        last_seen = MAX(contact_advert_paths.last_seen, excluded.last_seen),
+                        heard_count = contact_advert_paths.heard_count + excluded.heard_count
+                    """,
+                    (resolved_key, orphan_key),
+                )
+
+        if "contact_name_history" in child_tables:
+            await conn.execute(
+                "DELETE FROM contact_name_history WHERE public_key = ?",
+                (orphan_key,),
+            )
+        if "contact_advert_paths" in child_tables:
+            await conn.execute(
+                "DELETE FROM contact_advert_paths WHERE public_key = ?",
+                (orphan_key,),
+            )
+
+    await conn.commit()
@@ -0,0 +1,39 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add indexes used by the statistics endpoint's time-windowed scans."""
+    cursor = await conn.execute("SELECT name FROM sqlite_master WHERE type='table'")
+    tables = {row[0] for row in await cursor.fetchall()}
+
+    if "raw_packets" in tables:
+        cursor = await conn.execute("PRAGMA table_info(raw_packets)")
+        raw_packet_columns = {row[1] for row in await cursor.fetchall()}
+        if "timestamp" in raw_packet_columns:
+            await conn.execute(
+                "CREATE INDEX IF NOT EXISTS idx_raw_packets_timestamp ON raw_packets(timestamp)"
+            )
+
+    if "contacts" in tables:
+        cursor = await conn.execute("PRAGMA table_info(contacts)")
+        contact_columns = {row[1] for row in await cursor.fetchall()}
+        if {"type", "last_seen"}.issubset(contact_columns):
+            await conn.execute(
+                "CREATE INDEX IF NOT EXISTS idx_contacts_type_last_seen ON contacts(type, last_seen)"
+            )
+
+    if "messages" in tables:
+        cursor = await conn.execute("PRAGMA table_info(messages)")
+        message_columns = {row[1] for row in await cursor.fetchall()}
+        if {"type", "received_at", "conversation_key"}.issubset(message_columns):
+            await conn.execute(
+                """
+                CREATE INDEX IF NOT EXISTS idx_messages_type_received_conversation
+                ON messages(type, received_at, conversation_key)
+                """
+            )
+    await conn.commit()
@@ -0,0 +1,27 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add discovery_blocked_types column to app_settings.
+
+    Stores a JSON array of integer contact type codes (1=Client, 2=Repeater,
+    3=Room, 4=Sensor) whose advertisements should not create new contacts.
+    Empty list means all types are accepted.
+    """
+    try:
+        await conn.execute(
+            "ALTER TABLE app_settings ADD COLUMN discovery_blocked_types TEXT DEFAULT '[]'"
+        )
+    except Exception as e:
+        error_msg = str(e).lower()
+        if "duplicate column" in error_msg:
+            logger.debug("discovery_blocked_types column already exists, skipping")
+        elif "no such table" in error_msg:
+            logger.debug("app_settings table not ready, skipping discovery_blocked_types migration")
+        else:
+            raise
+    await conn.commit()
@@ -0,0 +1,158 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Rebuild FK tables with CASCADE/SET NULL and clean orphaned rows.
+
+    SQLite cannot ALTER existing FK constraints, so each table is rebuilt.
+    Orphaned child rows are cleaned up before the rebuild to ensure the
+    INSERT...SELECT into the new table (which has enforced FKs) succeeds.
+    """
+    import shutil
+    from pathlib import Path
+
+    # Back up the database before table rebuilds (skip for in-memory DBs).
+    cursor = await conn.execute("PRAGMA database_list")
+    db_row = await cursor.fetchone()
+    db_path = db_row[2] if db_row else ""
+    if db_path and db_path != ":memory:" and Path(db_path).exists():
+        backup_path = db_path + ".pre-fk-migration.bak"
+        for suffix in ("", "-wal", "-shm"):
+            src = Path(db_path + suffix)
+            if src.exists():
+                shutil.copy2(str(src), backup_path + suffix)
+        logger.info("Database backed up to %s before FK migration", backup_path)
+
+    # --- Phase 1: clean orphans (guard each table's existence) ---
+    tables_cursor = await conn.execute("SELECT name FROM sqlite_master WHERE type='table'")
+    existing_tables = {row[0] for row in await tables_cursor.fetchall()}
+
+    if "contact_advert_paths" in existing_tables and "contacts" in existing_tables:
+        await conn.execute(
+            "DELETE FROM contact_advert_paths "
+            "WHERE public_key NOT IN (SELECT public_key FROM contacts)"
+        )
+    if "contact_name_history" in existing_tables and "contacts" in existing_tables:
+        await conn.execute(
+            "DELETE FROM contact_name_history "
+            "WHERE public_key NOT IN (SELECT public_key FROM contacts)"
+        )
+    if "raw_packets" in existing_tables and "messages" in existing_tables:
+        # Guard: message_id column may not exist on very old schemas
+        col_cursor = await conn.execute("PRAGMA table_info(raw_packets)")
+        raw_cols = {row[1] for row in await col_cursor.fetchall()}
+        if "message_id" in raw_cols:
+            await conn.execute(
+                "UPDATE raw_packets SET message_id = NULL WHERE message_id IS NOT NULL "
+                "AND message_id NOT IN (SELECT id FROM messages)"
+            )
+    await conn.commit()
+    logger.debug("Cleaned orphaned child rows before FK rebuild")
+
+    # --- Phase 2: rebuild raw_packets with ON DELETE SET NULL ---
+    # Skip if raw_packets doesn't have message_id (pre-migration-18 schema)
+    raw_has_message_id = False
+    if "raw_packets" in existing_tables:
+        col_cursor2 = await conn.execute("PRAGMA table_info(raw_packets)")
+        raw_has_message_id = "message_id" in {row[1] for row in await col_cursor2.fetchall()}
+
+    if raw_has_message_id:
+        # Dynamically build column list based on what the old table actually has,
+        # since very old schemas may lack payload_hash (added in migration 28).
+        col_cursor3 = await conn.execute("PRAGMA table_info(raw_packets)")
+        old_cols = [row[1] for row in await col_cursor3.fetchall()]
+
+        new_col_defs = [
+            "id INTEGER PRIMARY KEY AUTOINCREMENT",
+            "timestamp INTEGER NOT NULL",
+            "data BLOB NOT NULL",
+            "message_id INTEGER",
+        ]
+        copy_cols = ["id", "timestamp", "data", "message_id"]
+        if "payload_hash" in old_cols:
+            new_col_defs.append("payload_hash BLOB")
+            copy_cols.append("payload_hash")
+        new_col_defs.append("FOREIGN KEY (message_id) REFERENCES messages(id) ON DELETE SET NULL")
+
+        cols_sql = ", ".join(new_col_defs)
+        copy_sql = ", ".join(copy_cols)
+        await conn.execute(f"CREATE TABLE raw_packets_fk ({cols_sql})")
+        await conn.execute(
+            f"INSERT INTO raw_packets_fk ({copy_sql}) SELECT {copy_sql} FROM raw_packets"
+        )
+        await conn.execute("DROP TABLE raw_packets")
+        await conn.execute("ALTER TABLE raw_packets_fk RENAME TO raw_packets")
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_raw_packets_message_id ON raw_packets(message_id)"
+        )
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_raw_packets_timestamp ON raw_packets(timestamp)"
+        )
+        if "payload_hash" in old_cols:
+            await conn.execute(
+                "CREATE UNIQUE INDEX IF NOT EXISTS idx_raw_packets_payload_hash ON raw_packets(payload_hash)"
+            )
+        await conn.commit()
+        logger.debug("Rebuilt raw_packets with ON DELETE SET NULL")
+
+    # --- Phase 3: rebuild contact_advert_paths with ON DELETE CASCADE ---
+    if "contact_advert_paths" in existing_tables:
+        await conn.execute(
+            """
+            CREATE TABLE contact_advert_paths_fk (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                public_key TEXT NOT NULL,
+                path_hex TEXT NOT NULL,
+                path_len INTEGER NOT NULL,
+                first_seen INTEGER NOT NULL,
+                last_seen INTEGER NOT NULL,
+                heard_count INTEGER NOT NULL DEFAULT 1,
+                UNIQUE(public_key, path_hex, path_len),
+                FOREIGN KEY (public_key) REFERENCES contacts(public_key) ON DELETE CASCADE
+            )
+            """
+        )
+        await conn.execute(
+            "INSERT INTO contact_advert_paths_fk (id, public_key, path_hex, path_len, first_seen, last_seen, heard_count) "
+            "SELECT id, public_key, path_hex, path_len, first_seen, last_seen, heard_count FROM contact_advert_paths"
+        )
+        await conn.execute("DROP TABLE contact_advert_paths")
+        await conn.execute("ALTER TABLE contact_advert_paths_fk RENAME TO contact_advert_paths")
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_contact_advert_paths_recent "
+            "ON contact_advert_paths(public_key, last_seen DESC)"
+        )
+        await conn.commit()
+        logger.debug("Rebuilt contact_advert_paths with ON DELETE CASCADE")
+
+    # --- Phase 4: rebuild contact_name_history with ON DELETE CASCADE ---
+    if "contact_name_history" in existing_tables:
+        await conn.execute(
+            """
+            CREATE TABLE contact_name_history_fk (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                public_key TEXT NOT NULL,
+                name TEXT NOT NULL,
+                first_seen INTEGER NOT NULL,
+                last_seen INTEGER NOT NULL,
+                UNIQUE(public_key, name),
+                FOREIGN KEY (public_key) REFERENCES contacts(public_key) ON DELETE CASCADE
+            )
+            """
+        )
+        await conn.execute(
+            "INSERT INTO contact_name_history_fk (id, public_key, name, first_seen, last_seen) "
+            "SELECT id, public_key, name, first_seen, last_seen FROM contact_name_history"
+        )
+        await conn.execute("DROP TABLE contact_name_history")
+        await conn.execute("ALTER TABLE contact_name_history_fk RENAME TO contact_name_history")
+        await conn.execute(
+            "CREATE INDEX IF NOT EXISTS idx_contact_name_history_key "
+            "ON contact_name_history(public_key, last_seen DESC)"
+        )
+        await conn.commit()
+        logger.debug("Rebuilt contact_name_history with ON DELETE CASCADE")
@@ -0,0 +1,27 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Create repeater_telemetry_history table for JSON-blob telemetry snapshots."""
+    await conn.execute(
+        """
+        CREATE TABLE IF NOT EXISTS repeater_telemetry_history (
+            id INTEGER PRIMARY KEY AUTOINCREMENT,
+            public_key TEXT NOT NULL,
+            timestamp INTEGER NOT NULL,
+            data TEXT NOT NULL,
+            FOREIGN KEY (public_key) REFERENCES contacts(public_key) ON DELETE CASCADE
+        )
+        """
+    )
+    await conn.execute(
+        """
+        CREATE INDEX IF NOT EXISTS idx_repeater_telemetry_pk_ts
+            ON repeater_telemetry_history (public_key, timestamp)
+        """
+    )
+    await conn.commit()
@@ -0,0 +1,24 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Remove vestigial sidebar_sort_order column from app_settings."""
+    col_cursor = await conn.execute("PRAGMA table_info(app_settings)")
+    columns = {row[1] for row in await col_cursor.fetchall()}
+    if "sidebar_sort_order" in columns:
+        try:
+            await conn.execute("ALTER TABLE app_settings DROP COLUMN sidebar_sort_order")
+            await conn.commit()
+        except Exception as e:
+            error_msg = str(e).lower()
+            if "syntax error" in error_msg or "drop column" in error_msg:
+                logger.debug(
+                    "SQLite doesn't support DROP COLUMN, sidebar_sort_order column will remain"
+                )
+                await conn.commit()
+            else:
+                raise
@@ -0,0 +1,21 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add nullable per-channel path hash mode override column."""
+    tables_cursor = await conn.execute("SELECT name FROM sqlite_master WHERE type='table'")
+    if "channels" not in {row[0] for row in await tables_cursor.fetchall()}:
+        await conn.commit()
+        return
+    try:
+        await conn.execute("ALTER TABLE channels ADD COLUMN path_hash_mode_override INTEGER")
+        await conn.commit()
+    except Exception as e:
+        if "duplicate column" in str(e).lower():
+            await conn.commit()
+        else:
+            raise
@@ -0,0 +1,20 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add tracked_telemetry_repeaters JSON list column to app_settings."""
+    tables_cursor = await conn.execute("SELECT name FROM sqlite_master WHERE type='table'")
+    if "app_settings" not in {row[0] for row in await tables_cursor.fetchall()}:
+        await conn.commit()
+        return
+    col_cursor = await conn.execute("PRAGMA table_info(app_settings)")
+    columns = {row[1] for row in await col_cursor.fetchall()}
+    if "tracked_telemetry_repeaters" not in columns:
+        await conn.execute(
+            "ALTER TABLE app_settings ADD COLUMN tracked_telemetry_repeaters TEXT DEFAULT '[]'"
+        )
+        await conn.commit()
@@ -0,0 +1,20 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add auto_resend_channel boolean column to app_settings."""
+    tables_cursor = await conn.execute("SELECT name FROM sqlite_master WHERE type='table'")
+    if "app_settings" not in {row[0] for row in await tables_cursor.fetchall()}:
+        await conn.commit()
+        return
+    col_cursor = await conn.execute("PRAGMA table_info(app_settings)")
+    columns = {row[1] for row in await col_cursor.fetchall()}
+    if "auto_resend_channel" not in columns:
+        await conn.execute(
+            "ALTER TABLE app_settings ADD COLUMN auto_resend_channel INTEGER DEFAULT 0"
+        )
+        await conn.commit()
@@ -0,0 +1,93 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Move favorites from app_settings JSON blob to per-entity boolean columns.
+
+    1. Add ``favorite`` column to contacts and channels tables.
+    2. Backfill from the ``app_settings.favorites`` JSON array.
+    3. Drop the ``favorites`` column from app_settings.
+    """
+    import json as _json
+
+    # --- Add columns ---
+    tables_cursor = await conn.execute("SELECT name FROM sqlite_master WHERE type='table'")
+    existing_tables = {row[0] for row in await tables_cursor.fetchall()}
+    for table in ("contacts", "channels"):
+        if table not in existing_tables:
+            continue
+        col_cursor = await conn.execute(f"PRAGMA table_info({table})")
+        columns = {row[1] for row in await col_cursor.fetchall()}
+        if "favorite" not in columns:
+            await conn.execute(f"ALTER TABLE {table} ADD COLUMN favorite INTEGER DEFAULT 0")
+    await conn.commit()
+
+    # --- Backfill from JSON ---
+    tables_cursor = await conn.execute("SELECT name FROM sqlite_master WHERE type='table'")
+    if "app_settings" not in {row[0] for row in await tables_cursor.fetchall()}:
+        await conn.commit()
+        return
+
+    col_cursor = await conn.execute("PRAGMA table_info(app_settings)")
+    settings_columns = {row[1] for row in await col_cursor.fetchall()}
+    if "favorites" not in settings_columns:
+        await conn.commit()
+        return
+
+    cursor = await conn.execute("SELECT favorites FROM app_settings WHERE id = 1")
+    row = await cursor.fetchone()
+    if row and row[0]:
+        try:
+            favorites = _json.loads(row[0])
+        except (ValueError, TypeError):
+            favorites = []
+
+        contact_keys = []
+        channel_keys = []
+        for fav in favorites:
+            if not isinstance(fav, dict):
+                continue
+            fav_type = fav.get("type")
+            fav_id = fav.get("id")
+            if not fav_id:
+                continue
+            if fav_type == "contact":
+                contact_keys.append(fav_id)
+            elif fav_type == "channel":
+                channel_keys.append(fav_id)
+
+        if contact_keys:
+            placeholders = ",".join("?" for _ in contact_keys)
+            await conn.execute(
+                f"UPDATE contacts SET favorite = 1 WHERE public_key IN ({placeholders})",
+                contact_keys,
+            )
+        if channel_keys:
+            placeholders = ",".join("?" for _ in channel_keys)
+            await conn.execute(
+                f"UPDATE channels SET favorite = 1 WHERE key IN ({placeholders})",
+                channel_keys,
+            )
+        if contact_keys or channel_keys:
+            logger.info(
+                "Backfilled %d contact favorite(s) and %d channel favorite(s) from app_settings",
+                len(contact_keys),
+                len(channel_keys),
+            )
+    await conn.commit()
+
+    # --- Drop the JSON column ---
+    try:
+        await conn.execute("ALTER TABLE app_settings DROP COLUMN favorites")
+        await conn.commit()
+    except Exception as e:
+        error_msg = str(e).lower()
+        if "syntax error" in error_msg or "drop column" in error_msg:
+            logger.debug("SQLite doesn't support DROP COLUMN; favorites column will remain unused")
+            await conn.commit()
+        else:
+            raise
@@ -0,0 +1,43 @@
+import logging
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def migrate(conn: aiosqlite.Connection) -> None:
+    """Add sender_key to the incoming PRIV dedup index.
+
+    Room-server posts are stored as PRIV messages sharing one conversation_key
+    (the room contact).  Without sender_key in the uniqueness constraint, two
+    different room participants sending identical text in the same clock second
+    collide and the second message is silently dropped.
+
+    Adding COALESCE(sender_key, '') is strictly more permissive — no existing
+    rows can conflict — so the migration only needs to rebuild the index.
+    """
+    cursor = await conn.execute(
+        "SELECT name FROM sqlite_master WHERE type='table' AND name='messages'"
+    )
+    if await cursor.fetchone() is None:
+        await conn.commit()
+        return
+
+    # The index references type, conversation_key, sender_timestamp, outgoing,
+    # and sender_key.  Some migration tests create minimal messages tables that
+    # lack these columns.  Skip gracefully when the schema is too old.
+    col_cursor = await conn.execute("PRAGMA table_info(messages)")
+    columns = {row[1] for row in await col_cursor.fetchall()}
+    required = {"type", "conversation_key", "sender_timestamp", "outgoing", "sender_key"}
+    if not required.issubset(columns):
+        await conn.commit()
+        return
+
+    await conn.execute("DROP INDEX IF EXISTS idx_messages_incoming_priv_dedup")
+    await conn.execute(
+        """CREATE UNIQUE INDEX IF NOT EXISTS idx_messages_incoming_priv_dedup
+           ON messages(type, conversation_key, text, COALESCE(sender_timestamp, 0),
+                       COALESCE(sender_key, ''))
+           WHERE type = 'PRIV' AND outgoing = 0"""
+    )
+    await conn.commit()
@@ -0,0 +1,66 @@
+"""
+Database migrations using SQLite's user_version pragma.
+
+Migrations run automatically on startup. The user_version pragma tracks
+which migrations have been applied (defaults to 0 for existing databases).
+
+Each migration lives in its own file: ``_NNN_description.py``, exposing an
+``async def migrate(conn)`` entry point.  The runner auto-discovers files by
+numeric prefix and executes them in order.
+
+This approach is safe for existing users - their databases have user_version=0,
+so all migrations run in order on first startup after upgrade.
+"""
+
+import importlib
+import logging
+import pkgutil
+import re
+
+import aiosqlite
+
+logger = logging.getLogger(__name__)
+
+
+async def get_version(conn: aiosqlite.Connection) -> int:
+    """Get current schema version from SQLite user_version pragma."""
+    cursor = await conn.execute("PRAGMA user_version")
+    row = await cursor.fetchone()
+    return row[0] if row else 0
+
+
+async def set_version(conn: aiosqlite.Connection, version: int) -> None:
+    """Set schema version using SQLite user_version pragma."""
+    await conn.execute(f"PRAGMA user_version = {version}")
+
+
+async def run_migrations(conn: aiosqlite.Connection) -> int:
+    """
+    Run all pending migrations.
+
+    Returns the number of migrations applied.
+    """
+    version = await get_version(conn)
+    applied = 0
+
+    for module_info in sorted(pkgutil.iter_modules(__path__), key=lambda m: m.name):
+        match = re.match(r"_(\d+)_", module_info.name)
+        if not match:
+            continue
+        num = int(match.group(1))
+        if num <= version:
+            continue
+        logger.info("Applying migration %d: %s", num, module_info.name)
+        mod = importlib.import_module(f"{__name__}.{module_info.name}")
+        await mod.migrate(conn)
+        await set_version(conn, num)
+        applied += 1
+
+    if applied > 0:
+        logger.info(
+            "Applied %d migration(s), schema now at version %d", applied, await get_version(conn)
+        )
+    else:
+        logger.debug("Schema up to date at version %d", version)
+
+    return applied
@@ -4,6 +4,10 @@ from pydantic import BaseModel, Field

 from app.path_utils import normalize_contact_route, normalize_route_override

+# Valid MeshCore contact types: 0=unknown, 1=client, 2=repeater, 3=room, 4=sensor.
+# Corrupted radio data can produce values outside this range.
+_VALID_CONTACT_TYPES = frozenset({0, 1, 2, 3, 4})
+

 class ContactRoute(BaseModel):
    """A normalized contact route."""
@@ -59,16 +63,30 @@ class ContactUpsert(BaseModel):
                -1 if radio_data.get("out_path_len", -1) == -1 else 0,
            ),
        )
+        # Clamp invalid contact types to 0 (unknown) — corrupted radio data
+        # can produce values like 111 or 240 that break downstream branching.
+        raw_type = radio_data.get("type", 0)
+        contact_type = raw_type if raw_type in _VALID_CONTACT_TYPES else 0
+
+        # Null out impossible coordinates — the contact is still ingested,
+        # but garbage lat/lon (e.g. 1953.7) is discarded rather than stored.
+        lat = radio_data.get("adv_lat")
+        lon = radio_data.get("adv_lon")
+        if lat is not None and not (-90 <= lat <= 90):
+            lat = None
+        if lon is not None and not (-180 <= lon <= 180):
+            lon = None
+
        return cls(
            public_key=public_key,
            name=radio_data.get("adv_name"),
-            type=radio_data.get("type", 0),
+            type=contact_type,
            flags=radio_data.get("flags", 0),
            direct_path=direct_path,
            direct_path_len=direct_path_len,
            direct_path_hash_mode=direct_path_hash_mode,
-            lat=radio_data.get("adv_lat"),
-            lon=radio_data.get("adv_lon"),
+            lat=lat,
+            lon=lon,
            last_advert=radio_data.get("last_advert"),
            on_radio=on_radio,
        )
@@ -91,6 +109,7 @@ class Contact(BaseModel):
    lon: float | None = None
    last_seen: int | None = None
    on_radio: bool = False
+    favorite: bool = False
    last_contacted: int | None = None  # Last time we sent/received a message
    last_read_at: int | None = None  # Server-side read state tracking
    first_seen: int | None = None
@@ -196,15 +215,6 @@ class Contact(BaseModel):
        """Convert the stored contact to the repository's write contract."""
        return ContactUpsert.from_contact(self, **changes)

-    @staticmethod
-    def from_radio_dict(public_key: str, radio_data: dict, on_radio: bool = False) -> dict:
-        """Backward-compatible dict wrapper over ContactUpsert.from_radio_dict()."""
-        return ContactUpsert.from_radio_dict(
-            public_key,
-            radio_data,
-            on_radio=on_radio,
-        ).model_dump()
-

 class CreateContactRequest(BaseModel):
    """Request to create a new contact."""
@@ -283,30 +293,6 @@ class NearestRepeater(BaseModel):
    heard_count: int


-class ContactDetail(BaseModel):
-    """Comprehensive contact profile data."""
-
-    contact: Contact
-    name_history: list[ContactNameHistory] = Field(default_factory=list)
-    dm_message_count: int = 0
-    channel_message_count: int = 0
-    most_active_rooms: list[ContactActiveRoom] = Field(default_factory=list)
-    advert_paths: list[ContactAdvertPath] = Field(default_factory=list)
-    advert_frequency: float | None = Field(
-        default=None,
-        description="Advert observations per hour (includes multi-path arrivals of same advert)",
-    )
-    nearest_repeaters: list[NearestRepeater] = Field(default_factory=list)
-
-
-class NameOnlyContactDetail(BaseModel):
-    """Channel activity summary for a sender name that is not tied to a known key."""
-
-    name: str
-    channel_message_count: int = 0
-    most_active_rooms: list[ContactActiveRoom] = Field(default_factory=list)
-
-
 class ContactAnalyticsHourlyBucket(BaseModel):
    """A single hourly activity bucket for contact analytics."""

@@ -354,7 +340,12 @@ class Channel(BaseModel):
        default=None,
        description="Per-channel outbound flood scope override (null = use global app setting)",
    )
+    path_hash_mode_override: int | None = Field(
+        default=None,
+        description="Per-channel path hash mode override (0=1-byte, 1=2-byte, 2=3-byte, null = use radio default)",
+    )
    last_read_at: int | None = None  # Server-side read state tracking
+    favorite: bool = False


 class ChannelMessageCounts(BaseModel):
@@ -375,6 +366,18 @@ class ChannelTopSender(BaseModel):
    message_count: int


+class PathHashWidthStats(BaseModel):
+    """Hop byte width distribution for parsed raw packets."""
+
+    total_packets: int = 0
+    single_byte: int = 0
+    double_byte: int = 0
+    triple_byte: int = 0
+    single_byte_pct: float = 0.0
+    double_byte_pct: float = 0.0
+    triple_byte_pct: float = 0.0
+
+
 class ChannelDetail(BaseModel):
    """Comprehensive channel profile data."""

@@ -383,6 +386,7 @@ class ChannelDetail(BaseModel):
    first_message_at: int | None = None
    unique_sender_count: int = 0
    top_senders_24h: list[ChannelTopSender] = Field(default_factory=list)
+    path_hash_width_24h: PathHashWidthStats = Field(default_factory=PathHashWidthStats)


 class MessagePath(BaseModel):
@@ -394,6 +398,8 @@ class MessagePath(BaseModel):
        default=None,
        description="Hop count. None = legacy (infer as len(path)//2, i.e. 1-byte hops)",
    )
+    rssi: int | None = Field(default=None, description="Last-hop RSSI in dBm")
+    snr: float | None = Field(default=None, description="Last-hop SNR in dB")


 class Message(BaseModel):
@@ -530,6 +536,9 @@ class RepeaterStatusResponse(BaseModel):
    flood_dups: int = Field(description="Duplicate flood packets")
    direct_dups: int = Field(description="Duplicate direct packets")
    full_events: int = Field(description="Full event queue count")
+    telemetry_history: list["TelemetryHistoryEntry"] = Field(
+        default_factory=list, description="Recent telemetry history snapshots"
+    )


 class RepeaterNodeInfoResponse(BaseModel):
@@ -767,13 +776,6 @@ class RadioDiscoveryResponse(BaseModel):
    )


-class Favorite(BaseModel):
-    """A favorite conversation."""
-
-    type: Literal["channel", "contact"] = Field(description="'channel' or 'contact'")
-    id: str = Field(description="Channel key or contact public key")
-
-
 class UnreadCounts(BaseModel):
    """Aggregated unread counts, mention flags, and last message times for all conversations."""

@@ -801,25 +803,14 @@ class AppSettings(BaseModel):
            "favorites reload first, then background fill targets about 80% of this value"
        ),
    )
-    favorites: list[Favorite] = Field(
-        default_factory=list, description="List of favorited conversations"
-    )
    auto_decrypt_dm_on_advert: bool = Field(
        default=True,
        description="Whether to attempt historical DM decryption on new contact advertisement",
    )
-    sidebar_sort_order: Literal["recent", "alpha"] = Field(
-        default="recent",
-        description="Sidebar sort order: 'recent' or 'alpha'",
-    )
    last_message_times: dict[str, int] = Field(
        default_factory=dict,
        description="Map of conversation state keys to last message timestamps",
    )
-    preferences_migrated: bool = Field(
-        default=False,
-        description="Whether preferences have been migrated from localStorage",
-    )
    advert_interval: int = Field(
        default=0,
        description="Periodic advertisement interval in seconds (0 = disabled)",
@@ -847,19 +838,17 @@ class AppSettings(BaseModel):
            "advertisements should not create new contacts; existing contacts are still updated"
        ),
    )
-
-
-class FanoutConfig(BaseModel):
-    """Configuration for a single fanout integration."""
-
-    id: str
-    type: str  # 'mqtt_private' | 'mqtt_community' | 'bot' | 'webhook' | 'apprise' | 'sqs'
-    name: str
-    enabled: bool
-    config: dict
-    scope: dict
-    sort_order: int = 0
-    created_at: int = 0
+    tracked_telemetry_repeaters: list[str] = Field(
+        default_factory=list,
+        description="Public keys of repeaters opted into periodic telemetry collection (max 8)",
+    )
+    auto_resend_channel: bool = Field(
+        default=False,
+        description=(
+            "When enabled, outgoing channel messages that receive no echo within 2 seconds "
+            "are automatically byte-perfect resent once (within the 30-second dedup window)"
+        ),
+    )


 class BusyChannel(BaseModel):
@@ -874,16 +863,6 @@ class ContactActivityCounts(BaseModel):
    last_week: int


-class PathHashWidthStats(BaseModel):
-    total_packets: int
-    single_byte: int
-    double_byte: int
-    triple_byte: int
-    single_byte_pct: float
-    double_byte_pct: float
-    triple_byte_pct: float
-
-
 class NoiseFloorSample(BaseModel):
    timestamp: int = Field(description="Unix timestamp of the sampled reading")
    noise_floor_dbm: int = Field(description="Noise floor in dBm")
@@ -898,13 +877,14 @@ class NoiseFloorHistoryStats(BaseModel):
    latest_timestamp: int | None = Field(
        default=None, description="Unix timestamp of the most recent sample"
    )
-    supported: bool | None = Field(
-        default=None,
-        description="Whether the connected radio appears to support radio stats sampling",
-    )
    samples: list[NoiseFloorSample] = Field(default_factory=list)


+class PacketsPerHourBucket(BaseModel):
+    timestamp: int = Field(description="Unix timestamp at the start of the hour")
+    count: int = Field(description="Number of packets received in that hour")
+
+
 class StatisticsResponse(BaseModel):
    busiest_channels_24h: list[BusyChannel]
    contact_count: int
@@ -920,4 +900,10 @@ class StatisticsResponse(BaseModel):
    repeaters_heard: ContactActivityCounts
    known_channels_active: ContactActivityCounts
    path_hash_width_24h: PathHashWidthStats
+    packets_per_hour_72h: list[PacketsPerHourBucket]
    noise_floor_24h: NoiseFloorHistoryStats
+
+
+class TelemetryHistoryEntry(BaseModel):
+    timestamp: int
+    data: dict
@@ -39,6 +39,7 @@ from app.repository import (
    ChannelRepository,
    ContactAdvertPathRepository,
    ContactRepository,
+    MessageRepository,
    RawPacketRepository,
 )
 from app.services.contact_reconciliation import (
@@ -68,6 +69,8 @@ async def create_message_from_decrypted(
    received_at: int | None = None,
    path: str | None = None,
    path_len: int | None = None,
+    rssi: int | None = None,
+    snr: float | None = None,
    channel_name: str | None = None,
    realtime: bool = True,
 ) -> int | None:
@@ -81,6 +84,8 @@ async def create_message_from_decrypted(
        received_at=received_at,
        path=path,
        path_len=path_len,
+        rssi=rssi,
+        snr=snr,
        channel_name=channel_name,
        realtime=realtime,
        broadcast_fn=broadcast_event,
@@ -95,6 +100,8 @@ async def create_dm_message_from_decrypted(
    received_at: int | None = None,
    path: str | None = None,
    path_len: int | None = None,
+    rssi: int | None = None,
+    snr: float | None = None,
    outgoing: bool = False,
    realtime: bool = True,
 ) -> int | None:
@@ -107,6 +114,8 @@ async def create_dm_message_from_decrypted(
        received_at=received_at,
        path=path,
        path_len=path_len,
+        rssi=rssi,
+        snr=snr,
        outgoing=outgoing,
        realtime=realtime,
        broadcast_fn=broadcast_event,
@@ -319,7 +328,9 @@ async def process_raw_packet(
    # deduplication in create_message_from_decrypted handles adding paths to existing messages.
    # This is more reliable than trying to look up the message via raw packet linking.
    if payload_type == PayloadType.GROUP_TEXT:
-        decrypt_result = await _process_group_text(raw_bytes, packet_id, ts, packet_info)
+        decrypt_result = await _process_group_text(
+            raw_bytes, packet_id, ts, packet_info, rssi=rssi, snr=snr
+        )
        if decrypt_result:
            result.update(decrypt_result)

@@ -330,7 +341,9 @@ async def process_raw_packet(

    elif payload_type == PayloadType.TEXT_MESSAGE:
        # Try to decrypt direct messages using stored private key and known contacts
-        decrypt_result = await _process_direct_message(raw_bytes, packet_id, ts, packet_info)
+        decrypt_result = await _process_direct_message(
+            raw_bytes, packet_id, ts, packet_info, rssi=rssi, snr=snr
+        )
        if decrypt_result:
            result.update(decrypt_result)

@@ -367,6 +380,8 @@ async def _process_group_text(
    packet_id: int,
    timestamp: int,
    packet_info: PacketInfo | None,
+    rssi: int | None = None,
+    snr: float | None = None,
 ) -> dict | None:
    """
    Process a GroupText (channel message) packet.
@@ -403,6 +418,8 @@ async def _process_group_text(
            received_at=timestamp,
            path=packet_info.path.hex() if packet_info else None,
            path_len=packet_info.path_length if packet_info else None,
+            rssi=rssi,
+            snr=snr,
        )

        return {
@@ -544,6 +561,8 @@ async def _process_direct_message(
    packet_id: int,
    timestamp: int,
    packet_info: PacketInfo | None,
+    rssi: int | None = None,
+    snr: float | None = None,
 ) -> dict | None:
    """
    Process a TEXT_MESSAGE (direct message) packet.
@@ -627,10 +646,30 @@ async def _process_direct_message(
        )

        if result is not None:
-            # Successfully decrypted!
+            # In the ambiguous direction case (both first bytes match), we
+            # defaulted to incoming.  Check if a matching outgoing message
+            # already exists — if so, this is actually our own outgoing echo
+            # and should be treated as such instead of creating a duplicate
+            # incoming row.
+            effective_outgoing = is_outgoing
+            if not is_outgoing and dest_hash == src_hash:
+                existing_outgoing = await MessageRepository.get_by_content(
+                    msg_type="PRIV",
+                    conversation_key=contact.public_key.lower(),
+                    text=result.message,
+                    sender_timestamp=result.timestamp,
+                    outgoing=True,
+                )
+                if existing_outgoing is not None:
+                    effective_outgoing = True
+                    logger.debug(
+                        "Ambiguous DM resolved as outgoing echo (matched existing sent msg %d)",
+                        existing_outgoing.id,
+                    )
+
            logger.debug(
                "Decrypted DM %s contact %s: %s",
-                "to" if is_outgoing else "from",
+                "to" if effective_outgoing else "from",
                contact.name or contact.public_key[:12],
                result.message[:50] if result.message else "",
            )
@@ -644,7 +683,9 @@ async def _process_direct_message(
                received_at=timestamp,
                path=packet_info.path.hex() if packet_info else None,
                path_len=packet_info.path_length if packet_info else None,
-                outgoing=is_outgoing,
+                rssi=rssi,
+                snr=snr,
+                outgoing=effective_outgoing,
            )

            return {
@@ -9,6 +9,7 @@ The path_len wire byte is packed as [hash_mode:2][hop_count:6]:
 Mode 3 (hash_size=4) is reserved and rejected.
 """

+from collections.abc import Iterable
 from dataclasses import dataclass

 MAX_PATH_SIZE = 64
@@ -244,3 +245,47 @@ def parse_explicit_hop_route(route_text: str) -> tuple[str, int, int]:
        raise ValueError(f"Explicit path exceeds MAX_PATH_SIZE={MAX_PATH_SIZE} bytes")

    return "".join(hops), len(hops), hash_size - 1
+
+
+def bucket_path_hash_widths(rows: Iterable) -> dict[str, int | float]:
+    """Bucket raw packet rows by hop hash width and return counts + percentages.
+
+    *rows* must be an already-fetched list whose elements have a ``data``
+    column containing raw packet bytes.
+    """
+    single_byte = 0
+    double_byte = 0
+    triple_byte = 0
+
+    for row in rows:
+        envelope = parse_packet_envelope(bytes(row["data"]))
+        if envelope is None:
+            continue
+        if envelope.hash_size == 1:
+            single_byte += 1
+        elif envelope.hash_size == 2:
+            double_byte += 1
+        elif envelope.hash_size == 3:
+            triple_byte += 1
+
+    total = single_byte + double_byte + triple_byte
+    if total == 0:
+        return {
+            "total_packets": 0,
+            "single_byte": 0,
+            "double_byte": 0,
+            "triple_byte": 0,
+            "single_byte_pct": 0.0,
+            "double_byte_pct": 0.0,
+            "triple_byte_pct": 0.0,
+        }
+
+    return {
+        "total_packets": total,
+        "single_byte": single_byte,
+        "double_byte": double_byte,
+        "triple_byte": triple_byte,
+        "single_byte_pct": (single_byte / total) * 100,
+        "double_byte_pct": (double_byte / total) * 100,
+        "triple_byte_pct": (triple_byte / total) * 100,
+    }
@@ -118,7 +118,7 @@ async def test_serial_device(port: str, baudrate: int, timeout: float = 3.0) ->
            return True

        return False
-    except asyncio.TimeoutError:
+    except TimeoutError:
        logger.debug("Device %s timed out", port)
        return False
    except Exception as e:
@@ -192,6 +192,9 @@ class RadioManager:
        if not blocking:
            if self._operation_lock.locked():
                raise RadioOperationBusyError(f"Radio is busy (operation: {name})")
+            # In single-threaded asyncio the lock cannot be acquired between the
+            # check above and the await below (no other coroutine runs until we
+            # yield). The await returns immediately for an uncontested lock.
            await self._operation_lock.acquire()
        else:
            await self._operation_lock.acquire()
@@ -21,13 +21,14 @@ from meshcore import EventType, MeshCore
 from app.channel_constants import PUBLIC_CHANNEL_KEY, PUBLIC_CHANNEL_NAME
 from app.config import settings
 from app.event_handlers import cleanup_expired_acks, on_contact_message
-from app.models import Contact, ContactUpsert
+from app.models import _VALID_CONTACT_TYPES, Contact, ContactUpsert
 from app.radio import RadioOperationBusyError
 from app.repository import (
    AmbiguousPublicKeyPrefixError,
    AppSettingsRepository,
    ChannelRepository,
    ContactRepository,
+    RepeaterTelemetryRepository,
 )
 from app.services.contact_reconciliation import (
    promote_prefix_contacts_for_contact,
@@ -155,6 +156,15 @@ ADVERT_CHECK_INTERVAL = 60
 # more frequently than this.
 MIN_ADVERT_INTERVAL = 3600

+# Periodic telemetry collection task handle
+_telemetry_collect_task: asyncio.Task | None = None
+
+# Telemetry collection interval (8 hours)
+TELEMETRY_COLLECT_INTERVAL = 8 * 3600
+
+# Initial delay before the first telemetry collection cycle (let radio settle)
+TELEMETRY_COLLECT_INITIAL_DELAY = 60
+
 # Counter to pause polling during repeater operations (supports nested pauses)
 _polling_pause_count: int = 0

@@ -253,70 +263,6 @@ async def should_run_full_periodic_sync(mc: MeshCore) -> bool:
    return False


-async def sync_and_offload_contacts(mc: MeshCore) -> dict:
-    """
-    Sync contacts from radio to database, then remove them from radio.
-    Returns counts of synced and removed contacts.
-    """
-    synced = 0
-    removed = 0
-
-    try:
-        # Get all contacts from radio
-        result = await mc.commands.get_contacts()
-
-        if result is None or result.type == EventType.ERROR:
-            logger.error(
-                "Failed to get contacts from radio: %s. "
-                "If you see this repeatedly, the radio may be visible on the "
-                "serial/TCP/BLE port but not responding to commands. Check for "
-                "another process with the serial port open (other RemoteTerm "
-                "instances, serial monitors, etc.), verify the firmware is "
-                "up-to-date and in client mode (not repeater), or try a "
-                "power cycle.",
-                result,
-            )
-            return {"synced": 0, "removed": 0, "error": str(result)}
-
-        contacts = result.payload or {}
-        logger.info("Found %d contacts on radio", len(contacts))
-
-        # Sync each contact to database, then remove from radio
-        for public_key, contact_data in contacts.items():
-            # Save to database
-            await ContactRepository.upsert(
-                ContactUpsert.from_radio_dict(public_key, contact_data, on_radio=False)
-            )
-            asyncio.create_task(
-                _reconcile_contact_messages_background(
-                    public_key,
-                    contact_data.get("adv_name"),
-                )
-            )
-            synced += 1
-
-            # Remove from radio
-            try:
-                remove_result = await mc.commands.remove_contact(contact_data)
-                if remove_result.type == EventType.OK:
-                    removed += 1
-                    _evict_removed_contact_from_library_cache(mc, public_key)
-                else:
-                    logger.warning(
-                        "Failed to remove contact %s: %s", public_key[:12], remove_result.payload
-                    )
-            except Exception as e:
-                logger.warning("Error removing contact %s: %s", public_key[:12], e)
-
-        logger.info("Synced %d contacts, removed %d from radio", synced, removed)
-
-    except Exception as e:
-        logger.error("Error during contact sync: %s", e)
-        return {"synced": synced, "removed": removed, "error": str(e)}
-
-    return {"synced": synced, "removed": removed}
-
-
 async def sync_and_offload_channels(mc: MeshCore, max_channels: int | None = None) -> dict:
    """
    Sync channels from radio to database, then clear them from radio.
@@ -361,7 +307,7 @@ async def sync_and_offload_channels(mc: MeshCore, max_channels: int | None = Non
            except Exception as e:
                logger.warning("Error clearing channel %d: %s", idx, e)

-        logger.info("Synced %d channels, cleared %d from radio", synced, cleared)
+        logger.debug("Synced %d channels, cleared %d from radio", synced, cleared)

    except Exception as e:
        logger.error("Error during channel sync: %s", e)
@@ -482,7 +428,6 @@ async def ensure_default_channels() -> None:

 async def sync_and_offload_all(mc: MeshCore) -> dict:
    """Run fast startup sync, then background contact reconcile."""
-    logger.info("Starting full radio sync and offload")

    # Contact on_radio is legacy/stale metadata. Clear it during the offload/reload
    # cycle so old rows stop claiming radio residency we do not actively track.
@@ -535,7 +480,7 @@ async def drain_pending_messages(mc: MeshCore) -> int:
            # Small delay between fetches
            await asyncio.sleep(0.1)

-        except asyncio.TimeoutError:
+        except TimeoutError:
            break
        except Exception as e:
            logger.warning("Error draining messages: %s", e, exc_info=True)
@@ -573,7 +518,7 @@ async def poll_for_messages(mc: MeshCore) -> int:
            # If we got a message, there might be more - drain them
            count += await drain_pending_messages(mc)

-    except asyncio.TimeoutError:
+    except TimeoutError:
        pass
    except Exception as e:
        logger.warning("Message poll exception: %s", e, exc_info=True)
@@ -998,10 +943,8 @@ async def sync_radio_time(mc: MeshCore) -> bool:
            except Exception:
                logger.warning("Reboot command failed", exc_info=True)
        elif _clock_reboot_attempted:
-            logger.warning(
-                "Clock skew persists after reboot — the radio likely has a "
-                "hardware RTC that preserved the wrong time.  A manual "
-                "'clkreboot' CLI command is needed to reset it."
+            logger.debug(
+                "Clock skew persists after reboot (hardware RTC); ignoring until next session."
            )

        return False
@@ -1111,7 +1054,7 @@ async def sync_contacts_from_radio(mc: MeshCore) -> dict:
            return {"synced": 0, "radio_contacts": {}, "error": str(result)}

        contacts = _normalize_radio_contacts_payload(result.payload)
-        logger.info("Found %d contacts on radio", len(contacts))
+        logger.debug("Found %d contacts on radio", len(contacts))

        for public_key, contact_data in contacts.items():
            await ContactRepository.upsert(
@@ -1125,7 +1068,29 @@ async def sync_contacts_from_radio(mc: MeshCore) -> dict:
            )
            synced += 1

-        logger.info("Synced %d contacts from radio snapshot", synced)
+        logger.debug("Synced %d contacts from radio snapshot", synced)
+
+        # Import radio-favorited contacts into app favorites.
+        # Only trust the favorite bit on contacts with a valid type (0-4);
+        # garbled radio data can have junk flags with bit 0 set.
+        radio_fav_keys = [
+            pk
+            for pk, data in contacts.items()
+            if data.get("flags", 0) & 0x01 and data.get("type", -1) in _VALID_CONTACT_TYPES
+        ]
+        if radio_fav_keys:
+            try:
+                imported = 0
+                for pk in radio_fav_keys:
+                    existing = await ContactRepository.get_by_key(pk)
+                    if existing and not existing.favorite:
+                        await ContactRepository.set_favorite(pk, True)
+                        imported += 1
+                if imported:
+                    logger.info("Imported %d radio favorite(s) into app favorites", imported)
+            except Exception as e:
+                logger.warning("Failed to import radio favorites: %s", e)
+
        return {"synced": synced, "radio_contacts": contacts}
    except Exception as e:
        logger.error("Error during contact snapshot sync: %s", e)
@@ -1337,26 +1302,9 @@ async def get_contacts_selected_for_radio_sync() -> list[Contact]:
    selected_contacts: list[Contact] = []
    selected_keys: set[str] = set()

+    # Favorites first — always loaded up to max_contacts
    favorite_contacts_loaded = 0
-    for favorite in app_settings.favorites:
-        if favorite.type != "contact":
-            continue
-        try:
-            contact = await ContactRepository.get_by_key_or_prefix(favorite.id)
-        except AmbiguousPublicKeyPrefixError:
-            logger.warning(
-                "Skipping favorite contact '%s': ambiguous key prefix; use full key",
-                favorite.id,
-            )
-            continue
-        if not contact:
-            continue
-        if len(contact.public_key) < 64:
-            logger.debug(
-                "Skipping unresolved prefix-only favorite contact '%s' for radio sync",
-                favorite.id,
-            )
-            continue
+    for contact in await ContactRepository.get_favorites():
        key = contact.public_key.lower()
        if key in selected_keys:
            continue
@@ -1588,3 +1536,203 @@ async def sync_recent_contacts_to_radio(force: bool = False, mc: MeshCore | None
    except Exception as e:
        logger.error("Error syncing contacts to radio: %s", e, exc_info=True)
        return {"loaded": 0, "error": str(e)}
+
+
+# ---------------------------------------------------------------------------
+# Periodic repeater telemetry collection
+# ---------------------------------------------------------------------------
+
+
+async def _collect_repeater_telemetry(mc: MeshCore, contact: Contact) -> bool:
+    """Fetch status telemetry from a single repeater and record it.
+
+    Returns True on success, False on failure (logged, not raised).
+    """
+    try:
+        await mc.commands.add_contact(contact.to_radio_dict())
+        status = await mc.commands.req_status_sync(contact.public_key, timeout=10, min_timeout=5)
+    except Exception as e:
+        logger.debug(
+            "Telemetry collect: radio command failed for %s: %s",
+            contact.public_key[:12],
+            e,
+        )
+        return False
+
+    if status is None:
+        logger.debug("Telemetry collect: no response from %s", contact.public_key[:12])
+        return False
+
+    # Map to the same field names as the manual repeater status endpoint
+    data = {
+        "battery_volts": status.get("bat", 0) / 1000.0,
+        "tx_queue_len": status.get("tx_queue_len", 0),
+        "noise_floor_dbm": status.get("noise_floor", 0),
+        "last_rssi_dbm": status.get("last_rssi", 0),
+        "last_snr_db": status.get("last_snr", 0.0),
+        "packets_received": status.get("nb_recv", 0),
+        "packets_sent": status.get("nb_sent", 0),
+        "airtime_seconds": status.get("airtime", 0),
+        "rx_airtime_seconds": status.get("rx_airtime", 0),
+        "uptime_seconds": status.get("uptime", 0),
+        "sent_flood": status.get("sent_flood", 0),
+        "sent_direct": status.get("sent_direct", 0),
+        "recv_flood": status.get("recv_flood", 0),
+        "recv_direct": status.get("recv_direct", 0),
+        "flood_dups": status.get("flood_dups", 0),
+        "direct_dups": status.get("direct_dups", 0),
+        "full_events": status.get("full_evts", 0),
+    }
+
+    # Best-effort LPP sensor fetch — failure here does not fail the overall
+    # collection; status telemetry is still recorded without sensor data.
+    try:
+        lpp_raw = await mc.commands.req_telemetry_sync(
+            contact.public_key, timeout=10, min_timeout=5
+        )
+        if lpp_raw:
+            lpp_sensors = []
+            for entry in lpp_raw:
+                value = entry.get("value", 0)
+                # Skip multi-value sensors (GPS, accelerometer, etc.)
+                if isinstance(value, dict):
+                    continue
+                lpp_sensors.append(
+                    {
+                        "channel": entry.get("channel", 0),
+                        "type_name": str(entry.get("type", "unknown")),
+                        "value": value,
+                    }
+                )
+            if lpp_sensors:
+                data["lpp_sensors"] = lpp_sensors
+    except Exception as e:
+        logger.debug(
+            "Telemetry collect: LPP sensor fetch failed for %s (non-fatal): %s",
+            contact.public_key[:12],
+            e,
+        )
+
+    try:
+        timestamp = int(time.time())
+        await RepeaterTelemetryRepository.record(
+            public_key=contact.public_key,
+            timestamp=timestamp,
+            data=data,
+        )
+        logger.info(
+            "Telemetry collect: recorded snapshot for %s (%s)",
+            contact.name or contact.public_key[:12],
+            contact.public_key[:12],
+        )
+
+        # Dispatch to fanout modules (e.g. HA MQTT discovery)
+        from app.fanout.manager import fanout_manager
+
+        asyncio.create_task(
+            fanout_manager.broadcast_telemetry(
+                {
+                    "public_key": contact.public_key,
+                    "name": contact.name or contact.public_key[:12],
+                    "timestamp": timestamp,
+                    **data,
+                }
+            )
+        )
+
+        return True
+    except Exception as e:
+        logger.warning(
+            "Telemetry collect: failed to record for %s: %s",
+            contact.public_key[:12],
+            e,
+        )
+        return False
+
+
+async def _telemetry_collect_loop() -> None:
+    """Background task that collects telemetry from tracked repeaters every 8 hours.
+
+    Runs a first cycle after a short initial delay (so newly tracked repeaters
+    get a sample promptly), then sleeps the full interval between subsequent cycles.
+
+    Acquires the radio lock per-repeater (non-blocking) so manual operations can
+    interleave. Failures are logged and skipped.
+    """
+    first_run = True
+    while True:
+        try:
+            delay = TELEMETRY_COLLECT_INITIAL_DELAY if first_run else TELEMETRY_COLLECT_INTERVAL
+            await asyncio.sleep(delay)
+            first_run = False
+
+            if not radio_manager.is_connected:
+                logger.debug("Telemetry collect: radio not connected, skipping cycle")
+                continue
+
+            app_settings = await AppSettingsRepository.get()
+            tracked = app_settings.tracked_telemetry_repeaters
+            if not tracked:
+                continue
+
+            logger.info("Telemetry collect: starting cycle for %d repeater(s)", len(tracked))
+            collected = 0
+
+            for pub_key in tracked:
+                contact = await ContactRepository.get_by_key(pub_key)
+                if not contact or contact.type != 2:
+                    logger.debug(
+                        "Telemetry collect: skipping %s (not found or not repeater)",
+                        pub_key[:12],
+                    )
+                    continue
+
+                try:
+                    async with radio_manager.radio_operation(
+                        "telemetry_collect",
+                        blocking=False,
+                        suspend_auto_fetch=True,
+                    ) as mc:
+                        if await _collect_repeater_telemetry(mc, contact):
+                            collected += 1
+                except RadioOperationBusyError:
+                    logger.debug(
+                        "Telemetry collect: radio busy, skipping %s",
+                        pub_key[:12],
+                    )
+
+            logger.info(
+                "Telemetry collect: cycle complete, %d/%d successful",
+                collected,
+                len(tracked),
+            )
+
+        except asyncio.CancelledError:
+            logger.info("Telemetry collect task cancelled")
+            break
+        except Exception as e:
+            logger.error("Error in telemetry collect loop: %s", e, exc_info=True)
+
+
+def start_telemetry_collect() -> None:
+    """Start the periodic telemetry collection background task."""
+    global _telemetry_collect_task
+    if _telemetry_collect_task is None or _telemetry_collect_task.done():
+        _telemetry_collect_task = asyncio.create_task(_telemetry_collect_loop())
+        logger.info(
+            "Started periodic telemetry collection (interval: %ds)",
+            TELEMETRY_COLLECT_INTERVAL,
+        )
+
+
+async def stop_telemetry_collect() -> None:
+    """Stop the periodic telemetry collection background task."""
+    global _telemetry_collect_task
+    if _telemetry_collect_task and not _telemetry_collect_task.done():
+        _telemetry_collect_task.cancel()
+        try:
+            await _telemetry_collect_task
+        except asyncio.CancelledError:
+            pass
+        _telemetry_collect_task = None
+        logger.info("Stopped periodic telemetry collection")
@@ -8,6 +8,7 @@ from app.repository.contacts import (
 from app.repository.fanout import FanoutConfigRepository
 from app.repository.messages import MessageRepository
 from app.repository.raw_packets import RawPacketRepository
+from app.repository.repeater_telemetry import RepeaterTelemetryRepository
 from app.repository.settings import AppSettingsRepository, StatisticsRepository

 __all__ = [
@@ -20,5 +21,6 @@ __all__ = [
    "FanoutConfigRepository",
    "MessageRepository",
    "RawPacketRepository",
+    "RepeaterTelemetryRepository",
    "StatisticsRepository",
 ]
@@ -26,7 +26,7 @@ class ChannelRepository:
        """Get a channel by its key (32-char hex string)."""
        cursor = await db.conn.execute(
            """
-            SELECT key, name, is_hashtag, on_radio, flood_scope_override, last_read_at
+            SELECT key, name, is_hashtag, on_radio, flood_scope_override, path_hash_mode_override, last_read_at, favorite
            FROM channels
            WHERE key = ?
            """,
@@ -40,7 +40,9 @@ class ChannelRepository:
                is_hashtag=bool(row["is_hashtag"]),
                on_radio=bool(row["on_radio"]),
                flood_scope_override=row["flood_scope_override"],
+                path_hash_mode_override=row["path_hash_mode_override"],
                last_read_at=row["last_read_at"],
+                favorite=bool(row["favorite"]),
            )
        return None

@@ -48,7 +50,7 @@ class ChannelRepository:
    async def get_all() -> list[Channel]:
        cursor = await db.conn.execute(
            """
-            SELECT key, name, is_hashtag, on_radio, flood_scope_override, last_read_at
+            SELECT key, name, is_hashtag, on_radio, flood_scope_override, path_hash_mode_override, last_read_at, favorite
            FROM channels
            ORDER BY name
            """
@@ -61,34 +63,22 @@ class ChannelRepository:
                is_hashtag=bool(row["is_hashtag"]),
                on_radio=bool(row["on_radio"]),
                flood_scope_override=row["flood_scope_override"],
+                path_hash_mode_override=row["path_hash_mode_override"],
                last_read_at=row["last_read_at"],
+                favorite=bool(row["favorite"]),
            )
            for row in rows
        ]

    @staticmethod
-    async def get_on_radio() -> list[Channel]:
-        """Return channels currently marked as resident on the radio in the database."""
+    async def set_favorite(key: str, value: bool) -> bool:
+        """Set or clear the favorite flag for a channel. Returns True if row was found."""
        cursor = await db.conn.execute(
-            """
-            SELECT key, name, is_hashtag, on_radio, flood_scope_override, last_read_at
-            FROM channels
-            WHERE on_radio = 1
-            ORDER BY name
-            """
+            "UPDATE channels SET favorite = ? WHERE key = ?",
+            (1 if value else 0, key.upper()),
        )
-        rows = await cursor.fetchall()
-        return [
-            Channel(
-                key=row["key"],
-                name=row["name"],
-                is_hashtag=bool(row["is_hashtag"]),
-                on_radio=bool(row["on_radio"]),
-                flood_scope_override=row["flood_scope_override"],
-                last_read_at=row["last_read_at"],
-            )
-            for row in rows
-        ]
+        await db.conn.commit()
+        return cursor.rowcount > 0

    @staticmethod
    async def delete(key: str) -> None:
@@ -123,6 +113,16 @@ class ChannelRepository:
        await db.conn.commit()
        return cursor.rowcount > 0

+    @staticmethod
+    async def update_path_hash_mode_override(key: str, path_hash_mode_override: int | None) -> bool:
+        """Set or clear a channel's path hash mode override."""
+        cursor = await db.conn.execute(
+            "UPDATE channels SET path_hash_mode_override = ? WHERE key = ?",
+            (path_hash_mode_override, key.upper()),
+        )
+        await db.conn.commit()
+        return cursor.rowcount > 0
+
    @staticmethod
    async def mark_all_read(timestamp: int) -> None:
        """Mark all channels as read at the given timestamp."""
@@ -170,6 +170,7 @@ class ContactRepository:
            lon=row["lon"],
            last_seen=row["last_seen"],
            on_radio=bool(row["on_radio"]),
+            favorite=bool(row["favorite"]) if "favorite" in available_columns else False,
            last_contacted=row["last_contacted"],
            last_read_at=row["last_read_at"],
            first_seen=row["first_seen"],
@@ -392,15 +393,30 @@ class ContactRepository:
            )
        await db.conn.commit()

+    @staticmethod
+    async def get_favorites() -> list[Contact]:
+        """Return all contacts marked as favorite."""
+        cursor = await db.conn.execute(
+            "SELECT * FROM contacts WHERE favorite = 1 AND LENGTH(public_key) = 64"
+        )
+        rows = await cursor.fetchall()
+        return [ContactRepository._row_to_contact(row) for row in rows]
+
+    @staticmethod
+    async def set_favorite(public_key: str, value: bool) -> None:
+        """Set or clear the favorite flag for a contact."""
+        await db.conn.execute(
+            "UPDATE contacts SET favorite = ? WHERE public_key = ?",
+            (1 if value else 0, public_key.lower()),
+        )
+        await db.conn.commit()
+
    @staticmethod
    async def delete(public_key: str) -> None:
        normalized = public_key.lower()
-        await db.conn.execute(
-            "DELETE FROM contact_name_history WHERE public_key = ?", (normalized,)
-        )
-        await db.conn.execute(
-            "DELETE FROM contact_advert_paths WHERE public_key = ?", (normalized,)
-        )
+        # contact_name_history and contact_advert_paths cascade via FK.
+        # Messages are intentionally preserved so history re-surfaces
+        # if the contact is re-added later.
        await db.conn.execute("DELETE FROM contacts WHERE public_key = ?", (normalized,))
        await db.conn.commit()

@@ -676,9 +692,18 @@ class ContactAdvertPathRepository:
        cursor = await db.conn.execute(
            """
            SELECT public_key, path_hex, path_len, first_seen, last_seen, heard_count
-            FROM contact_advert_paths
+            FROM (
+                SELECT *,
+                       ROW_NUMBER() OVER (
+                           PARTITION BY public_key
+                           ORDER BY last_seen DESC, heard_count DESC, path_len ASC, path_hex ASC
+                       ) AS rn
+                FROM contact_advert_paths
+            )
+            WHERE rn <= ?
            ORDER BY public_key ASC, last_seen DESC, heard_count DESC, path_len ASC, path_hex ASC
-            """
+            """,
+            (limit_per_contact,),
        )
        rows = await cursor.fetchall()

@@ -689,8 +714,6 @@ class ContactAdvertPathRepository:
            if paths is None:
                paths = []
                grouped[key] = paths
-            if len(paths) >= limit_per_contact:
-                continue
            paths.append(ContactAdvertPathRepository._row_to_path(row))

        return [
@@ -29,8 +29,7 @@ class MessageRepository:
    def _contact_activity_filter(public_key: str) -> tuple[str, list[Any]]:
        lower_key = public_key.lower()
        return (
-            "((type = 'PRIV' AND LOWER(conversation_key) = ?)"
-            " OR (type = 'CHAN' AND LOWER(sender_key) = ?))",
+            "((type = 'PRIV' AND conversation_key = ?) OR (type = 'CHAN' AND sender_key = ?))",
            [lower_key, lower_key],
        )

@@ -58,6 +57,8 @@ class MessageRepository:
        sender_timestamp: int | None = None,
        path: str | None = None,
        path_len: int | None = None,
+        rssi: int | None = None,
+        snr: float | None = None,
        txt_type: int = 0,
        signature: str | None = None,
        outgoing: bool = False,
@@ -79,8 +80,15 @@ class MessageRepository:
            entry: dict = {"path": path, "received_at": received_at}
            if path_len is not None:
                entry["path_len"] = path_len
+            if rssi is not None:
+                entry["rssi"] = rssi
+            if snr is not None:
+                entry["snr"] = snr
            paths_json = json.dumps([entry])

+        # Normalize sender_key to lowercase so queries can match without LOWER().
+        normalized_sender_key = sender_key.lower() if sender_key else sender_key
+
        cursor = await db.conn.execute(
            """
            INSERT OR IGNORE INTO messages (type, conversation_key, text, sender_timestamp,
@@ -99,7 +107,7 @@ class MessageRepository:
                signature,
                outgoing,
                sender_name,
-                sender_key,
+                normalized_sender_key,
            ),
        )
        await db.conn.commit()
@@ -114,6 +122,8 @@ class MessageRepository:
        path: str,
        received_at: int | None = None,
        path_len: int | None = None,
+        rssi: int | None = None,
+        snr: float | None = None,
    ) -> list[MessagePath]:
        """Add a new path to an existing message.

@@ -127,6 +137,10 @@ class MessageRepository:
        entry: dict = {"path": path, "received_at": ts}
        if path_len is not None:
            entry["path_len"] = path_len
+        if rssi is not None:
+            entry["rssi"] = rssi
+        if snr is not None:
+            entry["snr"] = snr
        new_entry = json.dumps(entry)
        await db.conn.execute(
            """UPDATE messages SET paths = json_insert(
@@ -259,10 +273,10 @@ class MessageRepository:

        if MessageRepository._looks_like_hex_prefix(value):
            if len(value) == 32:
-                clause += " OR UPPER(messages.conversation_key) = ?"
+                clause += " OR messages.conversation_key = ?"
                params.append(value.upper())
            else:
-                clause += " OR UPPER(messages.conversation_key) LIKE ? ESCAPE '\\'"
+                clause += " OR messages.conversation_key LIKE ? ESCAPE '\\'"
                params.append(f"{MessageRepository._escape_like(value.upper())}%")

        clause += "))"
@@ -281,13 +295,13 @@ class MessageRepository:
            priv_key_clause: str
            chan_key_clause: str
            if len(value) == 64:
-                priv_key_clause = "LOWER(messages.conversation_key) = ?"
-                chan_key_clause = "LOWER(sender_key) = ?"
+                priv_key_clause = "messages.conversation_key = ?"
+                chan_key_clause = "sender_key = ?"
                params.extend([lower_value, lower_value])
            else:
                escaped_prefix = f"{MessageRepository._escape_like(lower_value)}%"
-                priv_key_clause = "LOWER(messages.conversation_key) LIKE ? ESCAPE '\\'"
-                chan_key_clause = "LOWER(sender_key) LIKE ? ESCAPE '\\'"
+                priv_key_clause = "messages.conversation_key LIKE ? ESCAPE '\\'"
+                chan_key_clause = "sender_key LIKE ? ESCAPE '\\'"
                params.extend([escaped_prefix, escaped_prefix])

            clause += (
@@ -311,12 +325,12 @@ class MessageRepository:
        if blocked_keys:
            placeholders = ",".join("?" for _ in blocked_keys)
            blocked_matchers.append(
-                f"({prefix}type = 'PRIV' AND LOWER({prefix}conversation_key) IN ({placeholders}))"
+                f"({prefix}type = 'PRIV' AND {prefix}conversation_key IN ({placeholders}))"
            )
            params.extend(blocked_keys)
            blocked_matchers.append(
                f"({prefix}type = 'CHAN' AND {prefix}sender_key IS NOT NULL"
-                f" AND LOWER({prefix}sender_key) IN ({placeholders}))"
+                f" AND {prefix}sender_key IN ({placeholders}))"
            )
            params.extend(blocked_keys)

@@ -383,9 +397,9 @@ class MessageRepository:
        query = (
            f"SELECT {MessageRepository._message_select('messages')} FROM messages "
            "LEFT JOIN contacts ON messages.type = 'PRIV' "
-            "AND LOWER(messages.conversation_key) = LOWER(contacts.public_key) "
+            "AND messages.conversation_key = contacts.public_key "
            "LEFT JOIN channels ON messages.type = 'CHAN' "
-            "AND UPPER(messages.conversation_key) = UPPER(channels.key) "
+            "AND messages.conversation_key = channels.key "
            "WHERE 1=1"
        )
        params: list[Any] = []
@@ -543,10 +557,11 @@ class MessageRepository:
    @staticmethod
    async def increment_ack_count(message_id: int) -> int:
        """Increment ack count and return the new value."""
-        await db.conn.execute("UPDATE messages SET acked = acked + 1 WHERE id = ?", (message_id,))
-        await db.conn.commit()
-        cursor = await db.conn.execute("SELECT acked FROM messages WHERE id = ?", (message_id,))
+        cursor = await db.conn.execute(
+            "UPDATE messages SET acked = acked + 1 WHERE id = ? RETURNING acked", (message_id,)
+        )
        row = await cursor.fetchone()
+        await db.conn.commit()
        return row["acked"] if row else 1

    @staticmethod
@@ -673,7 +688,7 @@ class MessageRepository:
                           ELSE 0
                       END) > 0 as has_mention
            FROM messages m
-            JOIN contacts ct ON m.conversation_key = ct.public_key
+            LEFT JOIN contacts ct ON m.conversation_key = ct.public_key
            WHERE m.type = 'PRIV' AND m.outgoing = 0
              AND m.received_at > COALESCE(ct.last_read_at, 0)
              {blocked_sql}
@@ -784,12 +799,14 @@ class MessageRepository:

    @staticmethod
    async def get_channel_stats(conversation_key: str) -> dict:
-        """Get channel message statistics: time-windowed counts, first message, unique senders, top senders.
+        """Get channel message statistics: time-windowed counts, first message, unique senders, top senders, path hash widths.

-        Returns a dict with message_counts, first_message_at, unique_sender_count, top_senders_24h.
+        Returns a dict with message_counts, first_message_at, unique_sender_count, top_senders_24h, path_hash_width_24h.
        """
        import time as _time

+        from app.path_utils import bucket_path_hash_widths
+
        now = int(_time.time())
        t_1h = now - 3600
        t_24h = now - 86400
@@ -841,11 +858,25 @@ class MessageRepository:
            for r in top_rows
        ]

+        # Path hash width distribution for last 24h (in-Python parse of raw packet envelopes)
+        cursor3 = await db.conn.execute(
+            """
+            SELECT rp.data FROM raw_packets rp
+            JOIN messages m ON rp.message_id = m.id
+            WHERE m.type = 'CHAN' AND m.conversation_key = ?
+              AND rp.timestamp >= ?
+            """,
+            (conversation_key, t_24h),
+        )
+        rows3 = await cursor3.fetchall()
+        path_hash_width_24h = bucket_path_hash_widths(rows3)
+
        return {
            "message_counts": message_counts,
            "first_message_at": row["first_message_at"],
            "unique_sender_count": row["unique_sender_count"] or 0,
            "top_senders_24h": top_senders,
+            "path_hash_width_24h": path_hash_width_24h,
        }

    @staticmethod
@@ -1,5 +1,4 @@
 import logging
-import sqlite3
 import time
 from collections.abc import AsyncIterator
 from hashlib import sha256
@@ -35,46 +34,23 @@ class RawPacketRepository:
            # For malformed packets, hash the full data
            payload_hash = sha256(data).digest()

-        # Check if this payload already exists
+        cursor = await db.conn.execute(
+            "INSERT OR IGNORE INTO raw_packets (timestamp, data, payload_hash) VALUES (?, ?, ?)",
+            (ts, data, payload_hash),
+        )
+        await db.conn.commit()
+
+        if cursor.rowcount > 0:
+            assert cursor.lastrowid is not None
+            return (cursor.lastrowid, True)
+
+        # Duplicate payload — look up the existing row.
        cursor = await db.conn.execute(
            "SELECT id FROM raw_packets WHERE payload_hash = ?", (payload_hash,)
        )
        existing = await cursor.fetchone()
-
-        if existing:
-            # Duplicate - return existing packet ID
-            logger.debug(
-                "Duplicate payload detected (hash=%s..., existing_id=%d)",
-                payload_hash.hex()[:12],
-                existing["id"],
-            )
-            return (existing["id"], False)
-
-        # New packet - insert with hash
-        try:
-            cursor = await db.conn.execute(
-                "INSERT INTO raw_packets (timestamp, data, payload_hash) VALUES (?, ?, ?)",
-                (ts, data, payload_hash),
-            )
-            await db.conn.commit()
-            assert cursor.lastrowid is not None  # INSERT always returns a row ID
-            return (cursor.lastrowid, True)
-        except sqlite3.IntegrityError:
-            # Race condition: another insert with same payload_hash happened between
-            # our SELECT and INSERT. This is expected for duplicate packets arriving
-            # close together. Query again to get the existing ID.
-            logger.debug(
-                "Duplicate packet detected via race condition (payload_hash=%s), dropping",
-                payload_hash.hex()[:16],
-            )
-            cursor = await db.conn.execute(
-                "SELECT id FROM raw_packets WHERE payload_hash = ?", (payload_hash,)
-            )
-            existing = await cursor.fetchone()
-            if existing:
-                return (existing["id"], False)
-            # This shouldn't happen, but if it does, re-raise
-            raise
+        assert existing is not None
+        return (existing["id"], False)

    @staticmethod
    async def get_undecrypted_count() -> int:
@@ -95,35 +71,55 @@ class RawPacketRepository:
        return row["oldest"] if row and row["oldest"] is not None else None

    @staticmethod
-    async def get_all_undecrypted() -> list[tuple[int, bytes, int]]:
-        """Get all undecrypted packets as (id, data, timestamp) tuples."""
-        cursor = await db.conn.execute(
-            "SELECT id, data, timestamp FROM raw_packets WHERE message_id IS NULL ORDER BY timestamp ASC"
-        )
-        rows = await cursor.fetchall()
-        return [(row["id"], bytes(row["data"]), row["timestamp"]) for row in rows]
+    async def stream_all_undecrypted(
+        batch_size: int = UNDECRYPTED_PACKET_BATCH_SIZE,
+    ) -> AsyncIterator[tuple[int, bytes, int]]:
+        """Yield all undecrypted packets as (id, data, timestamp) in bounded batches.
+
+        Uses keyset pagination so each batch is a fresh query with a fully
+        consumed cursor — no open statement held across yield boundaries.
+        """
+        last_id = -1
+        while True:
+            cursor = await db.conn.execute(
+                "SELECT id, data, timestamp FROM raw_packets "
+                "WHERE message_id IS NULL AND id > ? ORDER BY id ASC LIMIT ?",
+                (last_id, batch_size),
+            )
+            rows = await cursor.fetchall()
+            await cursor.close()
+            if not rows:
+                break
+            for row in rows:
+                last_id = row["id"]
+                yield (row["id"], bytes(row["data"]), row["timestamp"])

    @staticmethod
    async def stream_undecrypted_text_messages(
        batch_size: int = UNDECRYPTED_PACKET_BATCH_SIZE,
    ) -> AsyncIterator[tuple[int, bytes, int]]:
-        """Yield undecrypted TEXT_MESSAGE packets in bounded-size batches."""
-        cursor = await db.conn.execute(
-            "SELECT id, data, timestamp FROM raw_packets WHERE message_id IS NULL ORDER BY timestamp ASC"
-        )
-        try:
-            while True:
-                rows = await cursor.fetchmany(batch_size)
-                if not rows:
-                    break
+        """Yield undecrypted TEXT_MESSAGE packets in bounded-size batches.

-                for row in rows:
-                    data = bytes(row["data"])
-                    payload_type = get_packet_payload_type(data)
-                    if payload_type == PayloadType.TEXT_MESSAGE:
-                        yield (row["id"], data, row["timestamp"])
-        finally:
+        Uses keyset pagination so each batch is a fresh query with a fully
+        consumed cursor — no open statement held across yield boundaries.
+        """
+        last_id = -1
+        while True:
+            cursor = await db.conn.execute(
+                "SELECT id, data, timestamp FROM raw_packets "
+                "WHERE message_id IS NULL AND id > ? ORDER BY id ASC LIMIT ?",
+                (last_id, batch_size),
+            )
+            rows = await cursor.fetchall()
            await cursor.close()
+            if not rows:
+                break
+            for row in rows:
+                last_id = row["id"]
+                data = bytes(row["data"])
+                payload_type = get_packet_payload_type(data)
+                if payload_type == PayloadType.TEXT_MESSAGE:
+                    yield (row["id"], data, row["timestamp"])

    @staticmethod
    async def count_undecrypted_text_messages(
@@ -187,12 +183,3 @@ class RawPacketRepository:
        cursor = await db.conn.execute("DELETE FROM raw_packets WHERE message_id IS NOT NULL")
        await db.conn.commit()
        return cursor.rowcount
-
-    @staticmethod
-    async def get_undecrypted_text_messages() -> list[tuple[int, bytes, int]]:
-        """Get all undecrypted TEXT_MESSAGE packets as (id, data, timestamp) tuples.
-
-        Filters raw packets to only include those with PayloadType.TEXT_MESSAGE (0x02).
-        These are direct messages that can be decrypted with contact ECDH keys.
-        """
-        return [packet async for packet in RawPacketRepository.stream_undecrypted_text_messages()]
@@ -0,0 +1,96 @@
+import json
+import logging
+import time
+
+from app.database import db
+
+logger = logging.getLogger(__name__)
+
+# Maximum age for telemetry history entries (30 days)
+_MAX_AGE_SECONDS = 30 * 86400
+
+# Maximum entries to keep per repeater (sanity cap)
+_MAX_ENTRIES_PER_REPEATER = 1000
+
+
+class RepeaterTelemetryRepository:
+    @staticmethod
+    async def record(
+        public_key: str,
+        timestamp: int,
+        data: dict,
+    ) -> None:
+        """Insert a telemetry history row and prune stale entries."""
+        await db.conn.execute(
+            """
+            INSERT INTO repeater_telemetry_history
+                (public_key, timestamp, data)
+            VALUES (?, ?, ?)
+            """,
+            (public_key, timestamp, json.dumps(data)),
+        )
+
+        # Prune entries older than 30 days
+        cutoff = int(time.time()) - _MAX_AGE_SECONDS
+        await db.conn.execute(
+            "DELETE FROM repeater_telemetry_history WHERE public_key = ? AND timestamp < ?",
+            (public_key, cutoff),
+        )
+
+        # Cap at _MAX_ENTRIES_PER_REPEATER (keep newest)
+        await db.conn.execute(
+            """
+            DELETE FROM repeater_telemetry_history
+            WHERE public_key = ? AND id NOT IN (
+                SELECT id FROM repeater_telemetry_history
+                WHERE public_key = ?
+                ORDER BY timestamp DESC
+                LIMIT ?
+            )
+            """,
+            (public_key, public_key, _MAX_ENTRIES_PER_REPEATER),
+        )
+
+        await db.conn.commit()
+
+    @staticmethod
+    async def get_history(public_key: str, since_timestamp: int) -> list[dict]:
+        """Return telemetry rows for a repeater since a given timestamp, ordered ASC."""
+        cursor = await db.conn.execute(
+            """
+            SELECT timestamp, data
+            FROM repeater_telemetry_history
+            WHERE public_key = ? AND timestamp >= ?
+            ORDER BY timestamp ASC
+            """,
+            (public_key, since_timestamp),
+        )
+        rows = await cursor.fetchall()
+        return [
+            {
+                "timestamp": row["timestamp"],
+                "data": json.loads(row["data"]),
+            }
+            for row in rows
+        ]
+
+    @staticmethod
+    async def get_latest(public_key: str) -> dict | None:
+        """Return the most recent telemetry row for a repeater, or None."""
+        cursor = await db.conn.execute(
+            """
+            SELECT timestamp, data
+            FROM repeater_telemetry_history
+            WHERE public_key = ?
+            ORDER BY timestamp DESC
+            LIMIT 1
+            """,
+            (public_key,),
+        )
+        row = await cursor.fetchone()
+        if row is None:
+            return None
+        return {
+            "timestamp": row["timestamp"],
+            "data": json.loads(row["data"]),
+        }
@@ -1,18 +1,18 @@
 import json
 import logging
 import time
-from typing import Any, Literal
+from typing import Any

 from app.database import db
-from app.models import AppSettings, Favorite
-from app.path_utils import parse_packet_envelope
+from app.models import AppSettings
+from app.path_utils import bucket_path_hash_widths

 logger = logging.getLogger(__name__)

 SECONDS_1H = 3600
 SECONDS_24H = 86400
+SECONDS_72H = 259200
 SECONDS_7D = 604800
-RAW_PACKET_STATS_BATCH_SIZE = 500


 class AppSettingsRepository:
@@ -26,10 +26,11 @@ class AppSettingsRepository:
        """
        cursor = await db.conn.execute(
            """
-            SELECT max_radio_contacts, favorites, auto_decrypt_dm_on_advert,
-                   sidebar_sort_order, last_message_times, preferences_migrated,
+            SELECT max_radio_contacts, auto_decrypt_dm_on_advert,
+                   last_message_times,
                   advert_interval, last_advert_time, flood_scope,
-                   blocked_keys, blocked_names, discovery_blocked_types
+                   blocked_keys, blocked_names, discovery_blocked_types,
+                   tracked_telemetry_repeaters, auto_resend_channel
            FROM app_settings WHERE id = 1
            """
        )
@@ -39,20 +40,6 @@ class AppSettingsRepository:
            # Should not happen after migration, but handle gracefully
            return AppSettings()

-        # Parse favorites JSON
-        favorites = []
-        if row["favorites"]:
-            try:
-                favorites_data = json.loads(row["favorites"])
-                favorites = [Favorite(**f) for f in favorites_data]
-            except (json.JSONDecodeError, TypeError, KeyError) as e:
-                logger.warning(
-                    "Failed to parse favorites JSON, using empty list: %s (data=%r)",
-                    e,
-                    row["favorites"][:100] if row["favorites"] else None,
-                )
-                favorites = []
-
        # Parse last_message_times JSON
        last_message_times: dict[str, int] = {}
        if row["last_message_times"]:
@@ -89,40 +76,48 @@ class AppSettingsRepository:
            except (json.JSONDecodeError, TypeError):
                discovery_blocked_types = []

-        # Validate sidebar_sort_order (fallback to "recent" if invalid)
-        sort_order = row["sidebar_sort_order"]
-        if sort_order not in ("recent", "alpha"):
-            sort_order = "recent"
+        # Parse tracked_telemetry_repeaters JSON
+        tracked_telemetry_repeaters: list[str] = []
+        try:
+            raw_tracked = row["tracked_telemetry_repeaters"]
+            if raw_tracked:
+                tracked_telemetry_repeaters = json.loads(raw_tracked)
+        except (json.JSONDecodeError, TypeError, KeyError):
+            tracked_telemetry_repeaters = []
+
+        # Parse auto_resend_channel boolean
+        try:
+            auto_resend_channel = bool(row["auto_resend_channel"])
+        except (KeyError, TypeError):
+            auto_resend_channel = False

        return AppSettings(
            max_radio_contacts=row["max_radio_contacts"],
-            favorites=favorites,
            auto_decrypt_dm_on_advert=bool(row["auto_decrypt_dm_on_advert"]),
-            sidebar_sort_order=sort_order,
            last_message_times=last_message_times,
-            preferences_migrated=bool(row["preferences_migrated"]),
            advert_interval=row["advert_interval"] or 0,
            last_advert_time=row["last_advert_time"] or 0,
            flood_scope=row["flood_scope"] or "",
            blocked_keys=blocked_keys,
            blocked_names=blocked_names,
            discovery_blocked_types=discovery_blocked_types,
+            tracked_telemetry_repeaters=tracked_telemetry_repeaters,
+            auto_resend_channel=auto_resend_channel,
        )

    @staticmethod
    async def update(
        max_radio_contacts: int | None = None,
-        favorites: list[Favorite] | None = None,
        auto_decrypt_dm_on_advert: bool | None = None,
-        sidebar_sort_order: str | None = None,
        last_message_times: dict[str, int] | None = None,
-        preferences_migrated: bool | None = None,
        advert_interval: int | None = None,
        last_advert_time: int | None = None,
        flood_scope: str | None = None,
        blocked_keys: list[str] | None = None,
        blocked_names: list[str] | None = None,
        discovery_blocked_types: list[int] | None = None,
+        tracked_telemetry_repeaters: list[str] | None = None,
+        auto_resend_channel: bool | None = None,
    ) -> AppSettings:
        """Update app settings. Only provided fields are updated."""
        updates = []
@@ -132,27 +127,14 @@ class AppSettingsRepository:
            updates.append("max_radio_contacts = ?")
            params.append(max_radio_contacts)

-        if favorites is not None:
-            updates.append("favorites = ?")
-            favorites_json = json.dumps([f.model_dump() for f in favorites])
-            params.append(favorites_json)
-
        if auto_decrypt_dm_on_advert is not None:
            updates.append("auto_decrypt_dm_on_advert = ?")
            params.append(1 if auto_decrypt_dm_on_advert else 0)

-        if sidebar_sort_order is not None:
-            updates.append("sidebar_sort_order = ?")
-            params.append(sidebar_sort_order)
-
        if last_message_times is not None:
            updates.append("last_message_times = ?")
            params.append(json.dumps(last_message_times))

-        if preferences_migrated is not None:
-            updates.append("preferences_migrated = ?")
-            params.append(1 if preferences_migrated else 0)
-
        if advert_interval is not None:
            updates.append("advert_interval = ?")
            params.append(advert_interval)
@@ -177,6 +159,14 @@ class AppSettingsRepository:
            updates.append("discovery_blocked_types = ?")
            params.append(json.dumps(discovery_blocked_types))

+        if tracked_telemetry_repeaters is not None:
+            updates.append("tracked_telemetry_repeaters = ?")
+            params.append(json.dumps(tracked_telemetry_repeaters))
+
+        if auto_resend_channel is not None:
+            updates.append("auto_resend_channel = ?")
+            params.append(1 if auto_resend_channel else 0)
+
        if updates:
            query = f"UPDATE app_settings SET {', '.join(updates)} WHERE id = 1"
            await db.conn.execute(query, params)
@@ -184,27 +174,6 @@ class AppSettingsRepository:

        return await AppSettingsRepository.get()

-    @staticmethod
-    async def add_favorite(fav_type: Literal["channel", "contact"], fav_id: str) -> AppSettings:
-        """Add a favorite, avoiding duplicates."""
-        settings = await AppSettingsRepository.get()
-
-        # Check if already favorited
-        if any(f.type == fav_type and f.id == fav_id for f in settings.favorites):
-            return settings
-
-        new_favorites = settings.favorites + [Favorite(type=fav_type, id=fav_id)]
-        return await AppSettingsRepository.update(favorites=new_favorites)
-
-    @staticmethod
-    async def remove_favorite(fav_type: Literal["channel", "contact"], fav_id: str) -> AppSettings:
-        """Remove a favorite."""
-        settings = await AppSettingsRepository.get()
-        new_favorites = [
-            f for f in settings.favorites if not (f.type == fav_type and f.id == fav_id)
-        ]
-        return await AppSettingsRepository.update(favorites=new_favorites)
-
    @staticmethod
    async def toggle_blocked_key(key: str) -> AppSettings:
        """Toggle a public key in the blocked list. Keys are normalized to lowercase."""
@@ -226,39 +195,6 @@ class AppSettingsRepository:
            new_names = settings.blocked_names + [name]
        return await AppSettingsRepository.update(blocked_names=new_names)

-    @staticmethod
-    async def migrate_preferences_from_frontend(
-        favorites: list[dict],
-        sort_order: str,
-        last_message_times: dict[str, int],
-    ) -> tuple[AppSettings, bool]:
-        """Migrate all preferences from frontend localStorage.
-
-        This is a one-time migration. If already migrated, returns current settings
-        without overwriting. Returns (settings, did_migrate) tuple.
-        """
-        settings = await AppSettingsRepository.get()
-
-        if settings.preferences_migrated:
-            # Already migrated, don't overwrite
-            return settings, False
-
-        # Convert frontend favorites format to Favorite objects
-        new_favorites = []
-        for f in favorites:
-            if f.get("type") in ("channel", "contact") and f.get("id"):
-                new_favorites.append(Favorite(type=f["type"], id=f["id"]))
-
-        # Update with migrated preferences and mark as migrated
-        settings = await AppSettingsRepository.update(
-            favorites=new_favorites,
-            sidebar_sort_order=sort_order if sort_order in ("recent", "alpha") else "recent",
-            last_message_times=last_message_times,
-            preferences_migrated=True,
-        )
-
-        return settings, True
-

 class StatisticsRepository:
    @staticmethod
@@ -338,6 +274,25 @@ class StatisticsRepository:
            "last_week": row["last_week"] or 0,
        }

+    @staticmethod
+    async def _packets_per_hour_72h() -> list[dict[str, int]]:
+        """Return packet counts bucketed by hour for the last 72 hours."""
+        now = int(time.time())
+        cutoff = now - SECONDS_72H
+        # Bucket timestamps to the start of each hour
+        cursor = await db.conn.execute(
+            """
+            SELECT (timestamp / 3600) * 3600 AS hour_ts, COUNT(*) AS count
+            FROM raw_packets
+            WHERE timestamp >= ?
+            GROUP BY hour_ts
+            ORDER BY hour_ts
+            """,
+            (cutoff,),
+        )
+        rows = await cursor.fetchall()
+        return [{"timestamp": row["hour_ts"], "count": row["count"]} for row in rows]
+
    @staticmethod
    async def _path_hash_width_24h() -> dict[str, int | float]:
        """Count parsed raw packets from the last 24h by hop hash width."""
@@ -346,48 +301,8 @@ class StatisticsRepository:
            "SELECT data FROM raw_packets WHERE timestamp >= ?",
            (now - SECONDS_24H,),
        )
-
-        single_byte = 0
-        double_byte = 0
-        triple_byte = 0
-
-        while True:
-            rows = await cursor.fetchmany(RAW_PACKET_STATS_BATCH_SIZE)
-            if not rows:
-                break
-
-            for row in rows:
-                envelope = parse_packet_envelope(bytes(row["data"]))
-                if envelope is None:
-                    continue
-                if envelope.hash_size == 1:
-                    single_byte += 1
-                elif envelope.hash_size == 2:
-                    double_byte += 1
-                elif envelope.hash_size == 3:
-                    triple_byte += 1
-
-        total_packets = single_byte + double_byte + triple_byte
-        if total_packets == 0:
-            return {
-                "total_packets": 0,
-                "single_byte": 0,
-                "double_byte": 0,
-                "triple_byte": 0,
-                "single_byte_pct": 0.0,
-                "double_byte_pct": 0.0,
-                "triple_byte_pct": 0.0,
-            }
-
-        return {
-            "total_packets": total_packets,
-            "single_byte": single_byte,
-            "double_byte": double_byte,
-            "triple_byte": triple_byte,
-            "single_byte_pct": (single_byte / total_packets) * 100,
-            "double_byte_pct": (double_byte / total_packets) * 100,
-            "triple_byte_pct": (triple_byte / total_packets) * 100,
-        }
+        rows = await cursor.fetchall()
+        return bucket_path_hash_widths(rows)

    @staticmethod
    async def get_all() -> dict:
@@ -455,6 +370,7 @@ class StatisticsRepository:
        repeaters_heard = await StatisticsRepository._activity_counts(contact_type=2)
        known_channels_active = await StatisticsRepository._known_channels_active()
        path_hash_width_24h = await StatisticsRepository._path_hash_width_24h()
+        packets_per_hour_72h = await StatisticsRepository._packets_per_hour_72h()

        return {
            "busiest_channels_24h": busiest_channels_24h,
@@ -471,4 +387,5 @@ class StatisticsRepository:
            "repeaters_heard": repeaters_heard,
            "known_channels_active": known_channels_active,
            "path_hash_width_24h": path_hash_width_24h,
+            "packets_per_hour_72h": packets_per_hour_72h,
        }
@@ -1,7 +1,8 @@
 import logging
+import re
 from hashlib import sha256

-from fastapi import APIRouter, HTTPException
+from fastapi import APIRouter, BackgroundTasks, HTTPException, Response, status
 from pydantic import BaseModel, Field

 from app.channel_constants import (
@@ -10,10 +11,12 @@ from app.channel_constants import (
    is_public_channel_key,
    is_public_channel_name,
 )
+from app.decoder import parse_packet, try_decrypt_packet_with_channel_key
 from app.models import Channel, ChannelDetail, ChannelMessageCounts, ChannelTopSender
+from app.packet_processor import create_message_from_decrypted
 from app.region_scope import normalize_region_scope
-from app.repository import ChannelRepository, MessageRepository
-from app.websocket import broadcast_event
+from app.repository import ChannelRepository, MessageRepository, RawPacketRepository
+from app.websocket import broadcast_event, broadcast_success

 logger = logging.getLogger(__name__)
 router = APIRouter(prefix="/channels", tags=["channels"])
@@ -31,12 +34,166 @@ class CreateChannelRequest(BaseModel):
    )


+class BulkCreateHashtagChannelsRequest(BaseModel):
+    channel_names: list[str] = Field(
+        min_length=1,
+        description="List of hashtag room names. Leading # is optional per entry.",
+    )
+    try_historical: bool = Field(
+        default=False,
+        description="Attempt one background historical decrypt sweep for the newly added rooms.",
+    )
+
+
+class BulkCreateHashtagChannelsResponse(BaseModel):
+    created_channels: list[Channel]
+    existing_count: int
+    invalid_names: list[str]
+    decrypt_started: bool = False
+    decrypt_total_packets: int = 0
+    message: str
+
+
 class ChannelFloodScopeOverrideRequest(BaseModel):
    flood_scope_override: str = Field(
        description="Blank clears the override; non-empty values temporarily override flood scope"
    )


+class ChannelPathHashModeOverrideRequest(BaseModel):
+    path_hash_mode_override: int | None = Field(
+        default=None,
+        ge=0,
+        le=2,
+        description="Path hash mode override (0=1-byte, 1=2-byte, 2=3-byte, null = use radio default)",
+    )
+
+
+def _derive_channel_identity(
+    requested_name: str,
+    request_key: str | None = None,
+) -> tuple[str, str, bool]:
+    is_hashtag = requested_name.startswith("#")
+
+    if is_public_channel_name(requested_name):
+        if request_key:
+            try:
+                key_bytes = bytes.fromhex(request_key)
+                if len(key_bytes) != 16:
+                    raise HTTPException(
+                        status_code=400,
+                        detail="Channel key must be exactly 16 bytes (32 hex chars)",
+                    )
+            except ValueError:
+                raise HTTPException(status_code=400, detail="Invalid hex string for key") from None
+            if key_bytes.hex().upper() != PUBLIC_CHANNEL_KEY:
+                raise HTTPException(
+                    status_code=400,
+                    detail=f'"{PUBLIC_CHANNEL_NAME}" must use the canonical Public key',
+                )
+        return PUBLIC_CHANNEL_KEY, PUBLIC_CHANNEL_NAME, False
+
+    if request_key and not is_hashtag:
+        try:
+            key_bytes = bytes.fromhex(request_key)
+            if len(key_bytes) != 16:
+                raise HTTPException(
+                    status_code=400, detail="Channel key must be exactly 16 bytes (32 hex chars)"
+                )
+        except ValueError:
+            raise HTTPException(status_code=400, detail="Invalid hex string for key") from None
+        key_hex = key_bytes.hex().upper()
+        if is_public_channel_key(key_hex):
+            raise HTTPException(
+                status_code=400,
+                detail=f'The canonical Public key may only be used for "{PUBLIC_CHANNEL_NAME}"',
+            )
+        return key_hex, requested_name, False
+
+    key_bytes = sha256(requested_name.encode("utf-8")).digest()[:16]
+    return key_bytes.hex().upper(), requested_name, is_hashtag
+
+
+def _normalize_bulk_hashtag_name(name: str) -> str | None:
+    trimmed = name.strip()
+    if not trimmed:
+        return None
+    normalized = trimmed.lstrip("#").strip()
+    if not normalized:
+        return None
+    if len(normalized) > 31:
+        return None
+    if not re.fullmatch(r"[A-Za-z0-9]+(?:-[A-Za-z0-9]+)*", normalized):
+        return None
+    return f"#{normalized}"
+
+
+async def _run_historical_channel_decryption_for_channels(
+    channels: list[tuple[bytes, str, str]],
+) -> None:
+    total = await RawPacketRepository.get_undecrypted_count()
+    decrypted_count = 0
+    matched_channel_names: set[str] = set()
+
+    if total == 0:
+        logger.info("No undecrypted packets to process for bulk channel decrypt")
+        return
+
+    logger.info(
+        "Starting bulk historical channel decryption of %d packets across %d channels",
+        total,
+        len(channels),
+    )
+
+    async for (
+        packet_id,
+        packet_data,
+        packet_timestamp,
+    ) in RawPacketRepository.stream_all_undecrypted():
+        packet_info = parse_packet(packet_data)
+        path_hex = packet_info.path.hex() if packet_info else None
+        path_len = packet_info.path_length if packet_info else None
+
+        for channel_key_bytes, channel_key_hex, channel_name in channels:
+            result = try_decrypt_packet_with_channel_key(packet_data, channel_key_bytes)
+            if result is None:
+                continue
+
+            msg_id = await create_message_from_decrypted(
+                packet_id=packet_id,
+                channel_key=channel_key_hex,
+                channel_name=channel_name,
+                sender=result.sender,
+                message_text=result.message,
+                timestamp=result.timestamp,
+                received_at=packet_timestamp,
+                path=path_hex,
+                path_len=path_len,
+                realtime=False,
+            )
+            if msg_id is not None:
+                decrypted_count += 1
+                matched_channel_names.add(channel_name)
+            break
+
+    logger.info(
+        "Bulk historical channel decryption complete: %d/%d packets decrypted across %d channels",
+        decrypted_count,
+        total,
+        len(matched_channel_names),
+    )
+
+    if decrypted_count > 0:
+        broadcast_success(
+            "Bulk historical decrypt complete",
+            (
+                f"Decrypted {decrypted_count} message{'s' if decrypted_count != 1 else ''} "
+                f"across {len(matched_channel_names)} room"
+                f"{'s' if len(matched_channel_names) != 1 else ''}"
+            ),
+        )
+
+
@router.get("", response_model=list[Channel])
 async def list_channels() -> list[Channel]:
    """List all channels from the database."""
@@ -58,6 +215,7 @@ async def get_channel_detail(key: str) -> ChannelDetail:
        first_message_at=stats["first_message_at"],
        unique_sender_count=stats["unique_sender_count"],
        top_senders_24h=[ChannelTopSender(**s) for s in stats["top_senders_24h"]],
+        path_hash_width_24h=stats["path_hash_width_24h"],
    )


@@ -69,50 +227,7 @@ async def create_channel(request: CreateChannelRequest) -> Channel:
    automatically when sending a message (see messages.py send_channel_message).
    """
    requested_name = request.name
-    is_hashtag = requested_name.startswith("#")
-
-    # Reserve the canonical Public channel so it cannot drift to another key,
-    # and the well-known Public key cannot be renamed to something else.
-    if is_public_channel_name(requested_name):
-        if request.key:
-            try:
-                key_bytes = bytes.fromhex(request.key)
-                if len(key_bytes) != 16:
-                    raise HTTPException(
-                        status_code=400,
-                        detail="Channel key must be exactly 16 bytes (32 hex chars)",
-                    )
-            except ValueError:
-                raise HTTPException(status_code=400, detail="Invalid hex string for key") from None
-            if key_bytes.hex().upper() != PUBLIC_CHANNEL_KEY:
-                raise HTTPException(
-                    status_code=400,
-                    detail=f'"{PUBLIC_CHANNEL_NAME}" must use the canonical Public key',
-                )
-        key_hex = PUBLIC_CHANNEL_KEY
-        channel_name = PUBLIC_CHANNEL_NAME
-        is_hashtag = False
-    elif request.key and not is_hashtag:
-        try:
-            key_bytes = bytes.fromhex(request.key)
-            if len(key_bytes) != 16:
-                raise HTTPException(
-                    status_code=400, detail="Channel key must be exactly 16 bytes (32 hex chars)"
-                )
-        except ValueError:
-            raise HTTPException(status_code=400, detail="Invalid hex string for key") from None
-        key_hex = key_bytes.hex().upper()
-        if is_public_channel_key(key_hex):
-            raise HTTPException(
-                status_code=400,
-                detail=f'The canonical Public key may only be used for "{PUBLIC_CHANNEL_NAME}"',
-            )
-        channel_name = requested_name
-    else:
-        # Derive key from name hash (same as meshcore library does)
-        key_bytes = sha256(requested_name.encode("utf-8")).digest()[:16]
-        key_hex = key_bytes.hex().upper()
-        channel_name = requested_name
+    key_hex, channel_name, is_hashtag = _derive_channel_identity(requested_name, request.key)

    logger.info("Creating channel %s: %s (hashtag=%s)", key_hex, channel_name, is_hashtag)

@@ -132,6 +247,81 @@ async def create_channel(request: CreateChannelRequest) -> Channel:
    return stored


+@router.post("/bulk-hashtag", response_model=BulkCreateHashtagChannelsResponse)
+async def bulk_create_hashtag_channels(
+    request: BulkCreateHashtagChannelsRequest,
+    background_tasks: BackgroundTasks,
+    response: Response,
+) -> BulkCreateHashtagChannelsResponse:
+    created_channels: list[Channel] = []
+    existing_count = 0
+    invalid_names: list[str] = []
+    decrypt_started = False
+    decrypt_total_packets = 0
+    decrypt_targets: list[tuple[bytes, str, str]] = []
+
+    for raw_name in request.channel_names:
+        normalized_name = _normalize_bulk_hashtag_name(raw_name)
+        if normalized_name is None:
+            invalid_names.append(raw_name)
+            continue
+
+        key_hex, channel_name, is_hashtag = _derive_channel_identity(normalized_name)
+        existing = await ChannelRepository.get_by_key(key_hex)
+        if existing is not None:
+            existing_count += 1
+            continue
+
+        await ChannelRepository.upsert(
+            key=key_hex,
+            name=channel_name,
+            is_hashtag=is_hashtag,
+            on_radio=False,
+        )
+        stored = await ChannelRepository.get_by_key(key_hex)
+        if stored is None:
+            raise HTTPException(
+                status_code=500,
+                detail="Channel was created but could not be reloaded",
+            )
+
+        created_channels.append(stored)
+        decrypt_targets.append((bytes.fromhex(stored.key), stored.key, stored.name))
+        _broadcast_channel_update(stored)
+
+    if request.try_historical and decrypt_targets:
+        decrypt_total_packets = await RawPacketRepository.get_undecrypted_count()
+        if decrypt_total_packets > 0:
+            background_tasks.add_task(
+                _run_historical_channel_decryption_for_channels, decrypt_targets
+            )
+            decrypt_started = True
+            response.status_code = status.HTTP_202_ACCEPTED
+
+    message = (
+        f"Created {len(created_channels)} room{'s' if len(created_channels) != 1 else ''}"
+        if created_channels
+        else "No new rooms were added"
+    )
+    if request.try_historical and decrypt_targets:
+        if decrypt_started:
+            message += (
+                f" and started background decrypt of {decrypt_total_packets} packet"
+                f"{'s' if decrypt_total_packets != 1 else ''}"
+            )
+        else:
+            message += "; no undecrypted packets were available"
+
+    return BulkCreateHashtagChannelsResponse(
+        created_channels=created_channels,
+        existing_count=existing_count,
+        invalid_names=invalid_names,
+        decrypt_started=decrypt_started,
+        decrypt_total_packets=decrypt_total_packets,
+        message=message,
+    )
+
+
@router.post("/{key}/mark-read")
 async def mark_channel_read(key: str) -> dict:
    """Mark a channel as read (update last_read_at timestamp)."""
@@ -168,6 +358,29 @@ async def set_channel_flood_scope_override(
    return refreshed


+@router.post("/{key}/path-hash-mode-override", response_model=Channel)
+async def set_channel_path_hash_mode_override(
+    key: str, request: ChannelPathHashModeOverrideRequest
+) -> Channel:
+    """Set or clear a per-channel path hash mode override."""
+    channel = await ChannelRepository.get_by_key(key)
+    if not channel:
+        raise HTTPException(status_code=404, detail="Channel not found")
+
+    updated = await ChannelRepository.update_path_hash_mode_override(
+        channel.key, request.path_hash_mode_override
+    )
+    if not updated:
+        raise HTTPException(status_code=500, detail="Failed to update path-hash-mode override")
+
+    refreshed = await ChannelRepository.get_by_key(channel.key)
+    if refreshed is None:
+        raise HTTPException(status_code=500, detail="Channel disappeared after update")
+
+    broadcast_event("channel", refreshed.model_dump())
+    return refreshed
+
+
@router.delete("/{key}")
 async def delete_channel(key: str) -> dict:
    """Delete a channel from the database by key.
@@ -8,7 +8,6 @@ from fastapi import APIRouter, BackgroundTasks, HTTPException, Query
 from meshcore import EventType
 from pydantic import BaseModel, Field

-from app.dependencies import require_connected
 from app.models import (
    Contact,
    ContactActiveRoom,
@@ -428,7 +427,7 @@ async def request_trace(public_key: str) -> TraceResponse:
    (no intermediate repeaters). This uses TRACE's dedicated width flags rather
    than the radio's normal path_hash_mode setting.
    """
-    require_connected()
+    radio_manager.require_connected()

    contact = await _resolve_contact_or_404(public_key)

@@ -487,7 +486,7 @@ async def request_trace(public_key: str) -> TraceResponse:
@router.post("/{public_key}/path-discovery", response_model=PathDiscoveryResponse)
 async def request_path_discovery(public_key: str) -> PathDiscoveryResponse:
    """Discover the current forward and return paths to a known contact."""
-    require_connected()
+    radio_manager.require_connected()

    contact = await _resolve_contact_or_404(public_key)
    pubkey_prefix = contact.public_key[:12]
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Jack Kingsman	4f54ec2c93	Updating changelog + build for 3.11.1	2026-04-12 20:50:12 -07:00
Jack Kingsman	eed38337c8	Add dummy SWer	2026-04-12 19:11:17 -07:00
Jack Kingsman	e1ee7fcd24	Add default precision	2026-04-12 18:59:44 -07:00
Jack Kingsman	2756b1ae8d	better wrapping around owner label on repeaters	2026-04-12 17:40:37 -07:00
Jack Kingsman	ef1d6a5a1a	Make all scripts +x	2026-04-12 17:35:54 -07:00
Jack Kingsman	14f42c59fe	Use localized units for repeater display	2026-04-12 17:32:07 -07:00
Jack Kingsman	b9414e84ee	Add LPP/tracked repeater telemetry and HA fanout	2026-04-12 17:23:25 -07:00
Jack Kingsman	95a17ca8ee	Merge pull request #174 from jkingsman/ha HomeAssistant MQTT Integration Module	2026-04-12 15:09:49 -07:00
Jack Kingsman	e6cedfbd0b	Improve db best practices. Contributes to fixing #179 .	2026-04-12 15:08:53 -07:00
Jack Kingsman	c3d0af1473	Fix memoization	2026-04-12 15:06:45 -07:00
Jack Kingsman	c24e291017	Destroy old discovery topics when the radio key changes	2026-04-12 14:59:41 -07:00
Jack Kingsman	d2d009ae79	Autoseed with radio identity	2026-04-12 14:54:36 -07:00
Jack Kingsman	d09166df84	HomeAssistant MQTT fanout	2026-04-12 14:36:13 -07:00
Jack Kingsman	f2762ab495	Merge pull request #178 from jkingsman/migration-updates Migration improvements	2026-04-12 14:35:26 -07:00
Jack Kingsman	a411562ca7	Filter keys to only search using prefix/beginning. Closes #180	2026-04-12 12:08:30 -07:00
Jack Kingsman	cde4d1744e	Fix async db handling. Closes #179 .	2026-04-12 11:57:37 -07:00
Jack Kingsman	4e73cd39c8	Migration improvements 2	2026-04-11 00:38:47 -07:00
Jack Kingsman	53b341d6fb	Make migrations more better	2026-04-10 16:28:03 -07:00
Jack Kingsman	76ac97010e	Use non-node20 checkout action	2026-04-10 16:19:21 -07:00
Jack Kingsman	53a4d8186a	Updating changelog + build for 3.11.0	2026-04-10 16:12:27 -07:00
Jack Kingsman	70e1669113	Improve test coverage	2026-04-10 16:04:02 -07:00
Jack Kingsman	3b1a292507	Docs updates and be consistent about node >=20	2026-04-10 15:57:47 -07:00
Jack Kingsman	4f19e1ec9a	Fix races and stale things	2026-04-10 15:54:03 -07:00
Jack Kingsman	59601bb98e	Assume that a same-second same-message same-first-byte-key DM is more likely an echo than them sending the same message, and multi-retry for flood scope restoration	2026-04-10 15:50:45 -07:00
Jack Kingsman	f6b0fd21fb	Don't consume DM resend attempt on busy radio	2026-04-10 15:46:19 -07:00
Jack Kingsman	8a4858a313	Don't consume DM resend attempt on busy radio	2026-04-10 15:44:50 -07:00
Jack Kingsman	442c2fad20	Fix some frontend display/quality/doc issues	2026-04-10 15:43:08 -07:00
Jack Kingsman	8cc542ce23	Fix same-second same-message collision in room servers with per-sender disambiguation at DB level	2026-04-10 15:36:53 -07:00
Jack Kingsman	a7258c120e	Merge pull request #177 from YourSandwich/feature/battery-status Add optional battery display to status bar	2026-04-10 14:55:39 -07:00
Jack Kingsman	8752320f52	Add some tests and move the helpers into their own TS file	2026-04-10 14:53:57 -07:00
Jack Kingsman	f9f046a05f	Fix inversion of const definition location	2026-04-10 14:51:19 -07:00
Jack Kingsman	390c0624ea	IIFE => memo for battery color/styling conversion	2026-04-10 14:49:05 -07:00
YourSandwich	2f55d11b0b	Add battery display toggles to Local Configuration	2026-04-10 23:38:29 +02:00
YourSandwich	fa0be24990	Add battery indicator to status bar	2026-04-10 23:38:29 +02:00
Jack Kingsman	1e22a21445	Add radio health &c. to fanout bus	2026-04-10 14:31:45 -07:00
YourSandwich	e09a3a01f7	Add localStorage helpers for battery display settings	2026-04-10 22:25:17 +02:00
Jack Kingsman	3bd756ee4e	Pluck in HA radio stats into the WS fanout endpoint	2026-04-10 12:39:37 -07:00
Jack Kingsman	43c5e0f67d	Improve e2e testing posture to make it sliiiightly less unfriendly for others to get working	2026-04-10 11:36:26 -07:00
Jack Kingsman	c0fc5fbba2	Add AUR download and test script	2026-04-10 11:30:05 -07:00
Jack Kingsman	c7248222dd	Updating changelog + build for 3.10.0	2026-04-10 11:16:16 -07:00
Jack Kingsman	1e18a91f12	Merge pull request #172 from YourSandwich/aur-install-instructions Add Arch Linux (AUR) packaging infrastructure	2026-04-10 10:54:49 -07:00
Jack Kingsman	18db6e4dd8	Make test script executable	2026-04-10 10:49:49 -07:00
Jack Kingsman	2393dadf1b	Unload the service on uninstall	2026-04-10 10:48:38 -07:00
Jack Kingsman	fd26576e0d	Use correct email	2026-04-10 10:47:21 -07:00
Sandwich	cb5a76eb5f	Replace manual user/group creation with sysusers.d and tmpfiles.d	2026-04-10 19:23:01 +02:00
Jack Kingsman	7f5dde119f	Update AGENTS.md	2026-04-10 00:15:57 -07:00
Jack Kingsman	799a721761	Be more defensive about systemd detection	2026-04-10 00:10:53 -07:00
Jack Kingsman	152a584f35	Fix TCP host	2026-04-10 00:10:41 -07:00
Jack Kingsman	5cc0476426	Fix port numbering	2026-04-10 00:06:22 -07:00
Jack Kingsman	e468c6c161	Change command palette shortcut	2026-04-09 23:45:16 -07:00
Jack Kingsman	e33537018b	Fix AUR username	2026-04-09 23:11:02 -07:00
Jack Kingsman	0727793560	Add test script	2026-04-09 23:08:32 -07:00
Jack Kingsman	5c4e04e024	Skip daemon reload if systemctl isn't around	2026-04-09 23:08:26 -07:00
Jack Kingsman	967269ef7d	Initial AUR work	2026-04-09 23:08:22 -07:00
Jack Kingsman	1903797d0d	Fix broken statistics pane e2e test	2026-04-09 22:30:12 -07:00
Jack Kingsman	bb5af5ba82	Bump apprise to 1.9.9. Closes #173 .	2026-04-09 17:20:57 -07:00
Sandwich	424da7e232	Add Arch Linux (AUR) install instructions to README Adds "Install Path 3: Arch Linux (AUR)" section covering both AUR helper and manual makepkg installation, linking to the published remoteterm-meshcore AUR package. Closes #171	2026-04-09 03:51:39 +02:00
Jack Kingsman	159df1ec5b	Revert "Add debug lines for fav click" This reverts commit `8e2e039985`.	2026-04-08 16:33:44 -07:00
Jack Kingsman	8e2e039985	Add debug lines for fav click	2026-04-08 16:18:46 -07:00
Jack Kingsman	01c86a486e	Add packet feed filters; closes #169 .	2026-04-08 14:44:41 -07:00
Jack Kingsman	7d5cfdec26	Add note about startup on windows	2026-04-07 22:07:31 -07:00
Jack Kingsman	5fe0ac0ad4	Be more memory concious on recent contact fetch	2026-04-07 16:41:34 -07:00
Jack Kingsman	b98102ccac	Add 72hr packet density view	2026-04-07 16:26:01 -07:00
Jack Kingsman	a02c3cae9e	Updating changelog + build for 3.9.0	2026-04-06 22:10:06 -07:00
Jack Kingsman	ca7349a1a8	Add autofocus to text boxes	2026-04-06 21:59:46 -07:00
Jack Kingsman	eeaa11b8b0	Fix lint bugs	2026-04-06 20:36:47 -07:00
Jack Kingsman	08eaf090b2	Be more guarded in the radio validity checks (and get outta here, you random repeaters I never favorited!)	2026-04-06 20:34:16 -07:00
Jack Kingsman	2f43420235	Add command palette	2026-04-06 20:27:55 -07:00
Jack Kingsman	af74663518	Add guard for favorites sync	2026-04-06 20:12:58 -07:00
Jack Kingsman	b7981c0450	Getting all Cal Raleigh up in here	2026-04-06 19:09:48 -07:00
Jack Kingsman	0f4976b9ee	Merge pull request #167 from jkingsman/migrate-favorites Add favorites as contact field (dug)	2026-04-05 22:19:01 -07:00
Jack Kingsman	1991f2515b	Support relative URLs. Closes #165 .	2026-04-05 22:11:12 -07:00
Jack Kingsman	a351c86ccb	Add favorites as contact field (dug)	2026-04-05 20:50:27 -07:00
Jack Kingsman	c2e1a3cbe6	Import radio favorites as favorites	2026-04-05 18:15:04 -07:00
jkingsman	c2d1339256	Default stale node pruning for visualizer to ON	2026-04-05 15:55:47 -07:00
jkingsman	cb7139a7e1	Always offer basic auth, move docker-not-found warning to the top	2026-04-05 15:41:02 -07:00
Jack Kingsman	6332387704	Define a better y domain for repeater battery voltage	2026-04-05 12:45:52 -07:00
Jack Kingsman	3f2b8e2a1f	Refocus CLI textbox after command completion. Closes #164 .	2026-04-05 11:55:52 -07:00
Jack Kingsman	40c37745b6	Massage the Readme a bit more	2026-04-05 11:55:31 -07:00
Jack Kingsman	9edac47aa2	Add clearer warning about RemoteTerm taking over the radio and governing contacts/channels loading. Closes #163 .	2026-04-05 11:49:57 -07:00
Jack Kingsman	44f8aafb66	Retain recent traces and make them click-to-trace. Closes #160 .	2026-04-04 16:43:12 -07:00
Jack Kingsman	9e3805f5d0	Use receipt time not sender time for display	2026-04-04 16:24:36 -07:00
Jack Kingsman	457799d8df	Calm down clock skew loggings	2026-04-04 15:31:30 -07:00
Jack Kingsman	de3ad2d51f	Calm it down on sync logs	2026-04-04 15:10:45 -07:00
Jack Kingsman	ad83bc7979	Show telemetry inline	2026-04-04 14:29:31 -07:00
Jack Kingsman	9ebf63491c	Have tests use prod regexes	2026-04-04 13:13:37 -07:00
Jack Kingsman	b19585db6d	Go crazy style on systemd escaping. Closes #159 .	2026-04-04 12:24:36 -07:00
Jack Kingsman	c28d22379e	Be a little gentler; call it a room finder rather than a cracker	2026-04-04 12:06:28 -07:00
Jack Kingsman	1e5ccf6c29	Add clearer issue identification for missing HTTPS context for channel finder	2026-04-04 12:03:07 -07:00
Jack Kingsman	81f5bde287	Add hop counts to width selection	2026-04-03 22:06:00 -07:00
Jack Kingsman	c33eb469ac	Updating changelog + build for 3.8.0	2026-04-03 19:36:27 -07:00
Jack Kingsman	0fe6584e7a	Add packet display to map & add map dark mode	2026-04-03 19:18:22 -07:00
Jack Kingsman	557d79d437	Add packets to general map	2026-04-03 18:57:34 -07:00
Jack Kingsman	daff3dcb4a	Drop low value tests	2026-04-03 17:55:02 -07:00
Jack Kingsman	77db7287d6	Drop lame imports	2026-04-03 17:51:26 -07:00
Jack Kingsman	67873e8dd9	Drop some duplicated logic and defns	2026-04-03 17:47:44 -07:00
Jack Kingsman	e2ddf5f79f	Move require connected down into the manager	2026-04-03 17:37:30 -07:00
Jack Kingsman	4a93641f04	Axe some dead code	2026-04-03 17:22:04 -07:00
Jack Kingsman	d5922a214b	Clear out old migration logic and replace with thin shim for favorites; sort order is lost	2026-04-03 17:15:41 -07:00
Jack Kingsman	7ad1ee26a4	Add RSSI/SNR to received messages. Closes #148 .	2026-04-03 15:20:44 -07:00
Jack Kingsman	08238aa464	Add close button to modal. Closes #156 (and modals lol), ish.	2026-04-03 14:54:59 -07:00
Jack Kingsman	1046baf741	Add auto-resend option for not-heard-repeated messages. Closes #154 .	2026-04-03 14:43:52 -07:00
Jack Kingsman	42e1b7b5d9	Add canonical style reference. Closes #155 .	2026-04-03 14:27:44 -07:00
Jack Kingsman	3ca4f7edf7	Fix missing test failures and patch double declared model	2026-04-03 14:15:19 -07:00
Jack Kingsman	55081d4a2d	Add hop width to channel info. Closes #153 .	2026-04-03 14:04:35 -07:00
Jack Kingsman	be2b2604df	Add intervalized repeater metrics collection. Closes #151 .	2026-04-03 13:45:39 -07:00
Jack Kingsman	35981d8f8b	Be more aggressive about resetting the hop width and warning if that doesn't work. This and the prior work closes #152 .	2026-04-03 13:16:43 -07:00
Jack Kingsman	8e998c03ba	Add channel path hash width override	2026-04-03 13:05:58 -07:00
Jack Kingsman	d802dd4212	Fix table display in primary agents.md	2026-04-02 20:31:54 -07:00
Jack Kingsman	7557eb1fa6	Merge pull request #150 from jkingsman/bugbash-v7 Bugbash v7	2026-04-02 20:20:23 -07:00
Jack Kingsman	6a4af5e602	More complete message lifecycle tests	2026-04-02 20:17:51 -07:00
Jack Kingsman	1895e6a919	Clean up legacy sort order	2026-04-02 20:16:16 -07:00
Jack Kingsman	975bf7f03f	Docs, dead code, and schema updates	2026-04-02 19:03:02 -07:00
Jack Kingsman	c7d5d3887d	Yield radio lock on build repeater ops and use INSERT OR IGNORE instead of check-then-act on packet ops	2026-04-02 18:53:34 -07:00
Jack Kingsman	5c93d8487e	Stop using db ops to do casing; unify on write and then our indices are happy once more	2026-04-02 18:50:56 -07:00
Jack Kingsman	5d2834a9fb	Add some tests around cascade deletion behaving now that we have FK pragma turned on	2026-04-02 18:46:37 -07:00
Jack Kingsman	cfe485bf29	Be kinder about streaming volume in memory	2026-04-02 18:43:48 -07:00
Jack Kingsman	e7f6bd0397	Bump python requirement so as not to hit toml issues	2026-04-02 18:41:03 -07:00
Jack Kingsman	1e7dc6af46	Don't clobber sort order	2026-04-02 18:40:25 -07:00
Jack Kingsman	af40cc3c8e	Add more recent screenshot	2026-04-02 18:06:29 -07:00
Jack Kingsman	2561b70fed	Fix tests for apprise redaction	2026-04-02 18:03:34 -07:00
Jack Kingsman	44f145b646	Updating changelog + build for 3.7.1	2026-04-02 18:01:22 -07:00
Jack Kingsman	55e2dc478d	Redact Apprise URLs	2026-04-02 17:59:41 -07:00
Jack Kingsman	0932800e1f	Fix lint	2026-04-02 17:38:35 -07:00
Jack Kingsman	c333eb25e3	Updating changelog + build for 3.7.0	2026-04-02 17:30:19 -07:00
Jack Kingsman	580aa1cefd	Correct TCP port	2026-04-02 13:55:05 -07:00
Jack Kingsman	30de09f71b	Merge pull request #126 from maplemesh/gnomeadrift/repeater_telemetry_history Logging battery voltage history from telemetry	2026-04-02 13:29:44 -07:00
Jack Kingsman	93d31adecd	Don't change historical migrations (cruft from rebasing) and don't overwrite data	2026-04-02 13:21:21 -07:00
Jack Kingsman	5f969017f7	Add some tests, make it an actual endpoint (whoops said we didn't need that) and tidy things up a bit	2026-04-02 12:43:42 -07:00
Gnome Adrift	967dd05fad	Prune telemetry entries, remove uplot comments, format code	2026-04-02 12:34:00 -07:00
Gnome Adrift	c808f0930b	Remove automatic telemetry querying, remove battery pane, add telemetry history pane	2026-04-02 12:31:51 -07:00
Gnome Adrift	87df4b4aa1	Fix for telemetry polling	2026-04-02 12:27:18 -07:00
Gnome Adrift	0511d6f69b	Make battery history update when fetching telemetry	2026-04-02 12:27:18 -07:00
Gnome Adrift	78b5598f67	First draft of repeater telemetry feature	2026-04-02 12:27:06 -07:00
Jack Kingsman	5e1bdb2cc1	Fix terminals on hash room parsing	2026-04-02 00:23:07 -07:00
Jack Kingsman	4420d44838	Add bulk room add	2026-04-02 00:19:25 -07:00
Jack Kingsman	ead1774cd3	Boost sidebar icon color	2026-04-01 22:17:04 -07:00
Jack Kingsman	0d45cbd849	Yolo to FK pragma Move to fk pragma	2026-04-01 22:09:00 -07:00