# Project status and handoff Last updated: 2026-09-09. ## Current direction Keep **Qwen 2.5**. The user explicitly canceled the earlier request to switch to a ChatGPT subscription/device-token login. Do not revive that migration because an old saved goal mentions it. Current work is the native iOS UI/UX redesign, restoring the music/AI server, and end-to-end qualification. The UI and playback changes are implemented, and the NAS deployment was verified on September 8. Full app qualification is **not complete**: physical-device tests and some broader live scenarios remain. This report summarizes prior evidence; server health and device availability must be rechecked when resuming. ## Completed work ### Native interface and usability - Reorganized the app into native **Talk, Music, Activity** tabs, sharing one session. - Replaced the large animated orb with a compact waveform/status presentation; added a clearer welcome screen, Voice/Type selection before connecting, explicit start/end controls, microphone meter, and recoverable connection errors. - Improved transcript readability, timestamps, live updates, copy/selection, follow-latest scrolling, and explicit Cancel/Clear confirmation. - Reworked Settings around voice, input mode, model, and connection; server changes require an explicit validated save. Voice selection persists. - Improved Music library/search navigation, now-playing controls, seek/volume, empty/loading/error states, and validated station-entry sheet. - Improved Activity hierarchy and task detail presentation with an active-task badge. - Removed forced dark appearance; adopted adaptive system styling, Dynamic Type, and larger interaction targets, informed by the Impeccable native-iOS guidance. ### Functional fixes - Observed nested transcript/state/task objects so asynchronous updates render. - Preserved latest-wins server replacement; clear old media when switching servers. - Sent standalone media controls and reports to the default server room. - Refreshed track metadata as downloads become ready; exposed request/decoder errors. - Loaded artwork through pinned TLS for the NAS self-signed certificate. - Requested cached AAC (`?format=m4a`) for YouTube library tracks on iOS. - Corrected AVFoundation's content type to a UTI rather than an HTTP MIME string. - Changed the pinned media loader to deliver 256 KB byte ranges instead of buffering an entire song before delivering usable audio, addressing a reproduced timeout. ### Tests and tooling - Installed XcodeBuildMCP and its skills; used it for simulator build/run/test work. - Added the `VoiceUITests` target in `project.yml` and regenerated the Xcode project. - Added isolated DEBUG UI-test preferences and optional fixtures; normal application preferences are not reset by those UI-test launch arguments. - Added media-control/error/metadata tests and opt-in real-NAS UI tests. - Live tests retain failure screenshots/UI hierarchy and attempt to stop music they initiated even when an assertion fails. ## Backend deployment and cross-repository dependency SSH host: `root@192.168.86.2` (Unraid). App URL: `https://192.168.86.2:9178`. - `voice` image: `hope-voice-api-voice:68f4c46-ios-audio`, host networking, restart policy `unless-stopped`. Includes LiveKit, voice agent, token service, media API, and nginx. - LLM: `qwen2.5-vl-3b` through existing OpenWebUI at `http://192.168.86.2:3005/api`. Existing Azure speech configuration retained. - `voice`, `open-webui`, and `local-vl-models` were present and healthy when the user suspected deletion. Do not recreate healthy services unnecessarily. - Original library, certificates, and skills were retained under `/mnt/user/dev/hope-voice-api`; the dirty remote checkout was not overwritten. - Separate deployment source: `/mnt/user/dev/hope-voice-deploy-68f4c46`. - Stopped rollback container: `voice-before-ios-audio`. - Backend AAC conversion preserves original WebM files and serves cached AAC with byte-range support. Browser/raw requests retain their original behavior. **Important:** the backend patch and its test are in a different local repository: `/Users/shane/dev/hope-voice-api/web/media_server.py` and `/Users/shane/dev/hope-voice-api/tests/test_ios_audio.py`. They were still uncommitted at this handoff. Committing/pushing `voice-ios` does **not** publish those files. The deployed image contains the patch, but a future backend CI rebuild can lose it unless it is committed and deployed through that repository's workflow. Inspect that workflow before pushing: it may replace the live container. Launch configuration and rollback details: [DEPLOYMENT.md](DEPLOYMENT.md). Do not copy credentials or `.env` contents into documentation or commits. ## Verification performed Project `Voice.xcodeproj`, scheme `Voice_iOS`, iPhone 17 Pro simulator. - 68 unit tests passed; five offline UI tests passed (navigation/error recovery, settings persistence/validation, music validation, transcript/task UI, large text). - Real NAS tests verified Qwen replies, background task completion, reconnect, library loading, advancing playback time, pause, forward-seek control and stop. - Qwen/task/reconnect passed three consecutive iterations. - Music initially failed intermittently. Logs captured an AVPlayer timeout while buffering a whole 6 MB response. After bounded-range delivery, three consecutive live playback iterations passed. The earlier isolated agent-join failure remains unexplained; do not claim all intermittent connection issues are resolved. - Backend integration: 16 passed, 7 skipped, 2 deselected. - Backend AAC regression: one passed (conversion, cache reuse, original preservation, and byte ranges); deployed range request returned HTTP 206. - The final bounded-range change was verified by focused playback repetitions, not a new full combined regression run. Exact result-bundle names and qualifications: [VERIFICATION.md](VERIFICATION.md). Artifacts live on this Mac under the XcodeBuildMCP workspace, not in Git. ## What remains, in order 1. **Pair the physical iPhone.** Latest check detected it on USB and reported iOS 26.6.1, device ID `BF93D35E-0853-5470-A125-6EB0DB3F17AB`, state `unpaired`. The user had unlocked it but could not see it in Finder. Last guidance was to enable iOS devices in Finder's Sidebar settings and accept any accessory/Trust prompts. Recheck current state rather than repeating old cable assumptions. 2. **Build, sign, install and run on that iPhone.** No successful device install is recorded. Check Developer Mode, trust, signing team/profile, and LAN access. 3. **Qualify real audio:** microphone capture, TTS audibility, speaker/headphones, barge-in, media/voice handoff, Siri/phone interruptions, lock/background behavior, microphone-permission-denied fallback, and network-loss recovery. Some checks require the user's speech/listening or a real interruption; simulator state tests do not prove them. Follow [DEVICE_TESTING.md](DEVICE_TESTING.md). 4. **Broader regression:** rerun the full suite after the range-loader change; test longer playback, seek near EOF, slow network/cancellation and server replacement. Exercise new downloads, test-owned station add/play/remove, and voice/model selection against the live agent. Do not delete the user's existing library song. 5. **Final visual/accessibility pass:** inspect all populated/error screens in light and dark mode, large text and small-device layouts; test VoiceOver and focus. The existing large-text navigation test is not a full accessibility audit. 6. **Publish the backend compatibility source safely** in the sibling repository and ensure its normal deployment preserves mounts, certificates and Qwen settings. 7. Update this report with results and remaining limitations. Do not mark the whole app qualified until the outstanding end-to-end gates have evidence. ## Repository handoff The iOS work and this report belong to branch `feat/ios-app`, remote `origin` (`https://git.srmr.xyz/snowspeeder/hope-voice-ios.git`). The requested handoff commit includes the implementation, tests, deployment notes and startup instructions. Check Git history/remote status for the final commit and push result.