Mobile/Code tab: assistant replies arrive as one block after a long wait (no streaming)
Environment: Claude Code via the mobile app (Code tab), driving a headless Claude Code session running in tmux on a remote server (iPhone as frontend).
Problem: On mobile, the assistant's reply does not stream in incrementally. After sending a message, the user waits (often long, because the model does tool calls in the background), and then the entire response appears at once as one large block. For longer answers this feels like a big latency spike followed by a "wall of text" landing all at once — jarring on a phone.
Compared to: The terminal/desktop experience streams tokens as generated, so the wait feels productive and the length feels gradual.
Requests / ideas:
- Incremental streaming to the mobile/app frontend, like the terminal — even partial text as it's generated.
- If full streaming isn't feasible, a live progress indicator (e.g., "reading files… / thinking… / writing…") so the wait doesn't feel dead.
- Optionally surface tool-call progress so the user knows work is happening during the gap.
Impact: The combination of (a) no streaming + (b) long tool-driven latency makes the mobile Code tab feel unresponsive, especially for users driving a remote session from their phone.