- 1024px icon from web/brand/hope-icon.svg (forced-dark, grid-tuned)
- fix xcodegen: xcassets was never in the generated project (no
resources phase) — move it into sources with buildPhase: resources
- fix iconset manifest: idiom universal (universal-platform rejected
by actool), single 1024 universal/ios entry
- Transcript: selection suspension clears when press ends; Latest
re-enables follow fully (clears stale selection state)
- Send disabled until transport ready in pending Type composer
- Server URL: programmatic success-trim no longer erases its own
saved-confirmation; only user edits clear it
- Expose End conversation + readable draft during connecting/waiting
- Stop masking assistant speaking/thinking state in Type/muted modes
- Restrict failure diagnosis to real causes (no invented network text)
- Replace old blue tint with Rosewood everywhere
- Fix transcript short-content bottom-anchor (top-align short histories)
- Settings: visible Server URL label, focus on rejected save, edit clears
saved confirmation, trim-aware unchanged detection
- Task detail: live store lookup with missing-task state
- Task rows: status text label restored
TaskWorker resolved its model from LLM_MODEL env (frozen at MCP spawn),
so after a live set_model switch in the UI, dispatched background tasks
still asked for the old model. On local-vl-models (llama-swap
--models-max 1) that forces a swap against the loaded conversation
model; a force-kill mid-load surfaces as 400 'model failed to load'.
The agent now writes the selected model to <task_dir>/model.txt at
session start and on every switch; TaskWorker reads it with the env
value as fallback.
Move the media server, downloader, library volume, and browser player to the
standalone hope-media-music product. Keep Hope's music tools as a thin HTTP
client of the LAN service on port 8094.
Use a stable voice-room player scope instead of ephemeral LiveKit room names.
The agent now owns the conversation handshake: fresh reported playback mutes
its input, joining pauses music, and the last participant leaving resumes only
music it paused. Peer failures remain non-fatal.
Also move the voice deploy workflow onto the active Unraid Gitea runner and
remove yt-dlp/ffmpeg/media health and volume dependencies from the image.
The music feature is retired: delete the Media tab, MediaAPI,
MediaPlayerController, PinnedMediaResourceLoader, MediaPanelView, the
media_state control publish, media scope handling in VoiceSession, the
volume setting, and all media tests. App is now Talk + Activity tabs.
Backend media API in hope-voice-api is untouched (still serves the web UI).
- MediaAPI client for all 10 media endpoints + Codable models
- PinnedMediaResourceLoader: AVAssetResourceLoaderDelegate proxying
https media through the pinned-TLS session (AVPlayer ignores registered
URLProtocol subclasses and can't trust the self-signed cert)
- MediaPlayerController: AVPlayer + 1.5s poll reconciliation ported from
web app.js (track swap, late audio_url, desired state, seek_hint,
reporting, enter-media-scope carry-over, post-talk resume)
- MediaPanelView UI + music.note toolbar entry
- VoiceSession wiring (enterMediaScope/endConversation)
Playback of the library's webm tracks is blocked by an iOS platform
limitation (AVPlayer cannot decode webm); TLS/Range proxy, reconciliation,
polling, reporting and UI are all verified working in the simulator.
Background research tasks were producing garbage or failing because the
self-hosted Firecrawl /v1/scrape defaults to a ~30s timeout that's too short
for heavy news/research sites, and the small LLM won't reliably chain a
second scrape call after a search that only returns snippets.
task_worker.py:
- web_search now auto-fetches the top results directly from the container
(which has working outbound internet) and returns their real text, so the
model gets actual page content in one call instead of snippet echoes.
- web_scrape requests timeout: 90000 (Firecrawl renders heavy sites but needs
>30s), and falls back to a direct container fetch with tag-stripping if
Firecrawl times out or returns nothing.
task_registry.py / dispatch_mcp.py:
- Add task_status(id) and list_tasks() tools so Hope can check the live state
of any dispatched task (pending / running / completed / failed / STALLED).
- Track last_activity per task and surface a stall heuristic.
- Move the status summary into task_registry (no mcp dependency) so it's
unit-testable.
agent.py: prompt tells Hope about task_status / list_tasks.
tests: add task-status summary and last-activity regression tests.
Hope often starts an already-downloaded song from a voice command, with no
click on the player. Two things could leave it paused at 0:
- The <audio> element sits outside the LiveKit room's audio-unlock flow, so
autoplay policy rejected the poll-driven play(). Capture the first page
gesture (pointerdown/keydown/touchstart) and play the current src inside it
to whitelist the element for autoplay.
- Player state is scoped per LiveKit room, but mediaScope() could return a
stale shell room after a reconnect. Derive it from the connected room.name
so the browser polls exactly the room the agent joined, and a "play X"
request can never land in a scope the browser isn't watching.
Add an in-container regression test (test_media_play_resolves_cached_ready_track)
pinning the server-side resolve: an already-ready track plays without a
re-download and serves a stable audio_url.
snowspeeder