Two magic keys let you work with a coding agent without typing everything on a phone keyboard: Images sends a picture to the agent, and Voice turns speech into terminal input. Voice is on the default pad, right of the Down arrow. Add Images, or bring either key back after deleting it, from the function list in the key chooser.
Send an image to a coding agent
- Attach to the tmux or Herdr session where Claude Code, Codex or another agent is running.
- Tap the Images key and choose Camera, Photo Library or Files.
- Warpeer uploads the image to the host over your SSH connection and pastes its full path at the agent's prompt.
- Add a sentence if you like, then send it yourself. Warpeer never presses Return for you.
Pasting an image from the iPhone clipboard into the terminal offers the same upload, so a screenshot you just copied reaches the host instead of a path that only exists on your phone.
Images can be up to 8 MiB. The host checks each upload against its digest before the path is pasted, and a failed or cancelled upload never leaves a partial file in place of an earlier one. If you switch pane, session or connection while an upload is running, nothing is pasted and the image stays selected so you can retry.
Warpeer pastes a plain, quoted path. Whether the agent reads the image depends on the agent and its model.
Dictate terminal input
- Tap the Voice key. The first time, allow microphone and speech recognition access.
- Speak. A live waveform shows that Warpeer is listening, and the transcript appears as you talk.
- Pause and continue as often as you need, and edit the transcript by hand.
- Tap Insert Text to place the text on the current shell line or agent prompt. It is not run until you press Return.
Dictation uses Apple Speech in your preferred language. Apple may process the audio on its servers under Apple's privacy policy. Images are sent only to the host you are connected to.