Photo and voice
Termivoa lets you send a photo to a coding agent, or speak a prompt, from your phone. Both are under compose when you have control of a session.
Send a photo
Section titled “Send a photo”- Tap Photo, then take or pick a photo.
- When the upload is done, the draft starts with
@.termivoa/uploads/<name>. This tells Claude Code and Codex to look at the file. - Add your prompt after it and tap Send ⏎.
✕ removes the photo from the message. The file stays: uploads are never deleted on their own.
What happens to the photo?
Section titled “What happens to the photo?”- The phone removes location (GPS) and other photo data (EXIF) before sending.
- The long side is scaled down to 2048 pixels at most.
- Only JPEG, PNG, WebP and GIF up to 5 MB are accepted.
- It is saved in the session’s project folder, in
.termivoa/uploads/. A.gitignorein.termivoa/keeps uploads out of git.
Speak a prompt
Section titled “Speak a prompt”- Tap Voice and allow the microphone if asked.
- Speak. Your words show as you talk.
- Tap Done · edit to add them to the draft, or Cancel to drop them.
- Check the text, then tap Send ⏎.
Voice never sends on its own. Listening stops when you close the panel or lock the phone.
Where does my voice go?
Section titled “Where does my voice go?”Voice uses your browser’s speech service. On iPhone, Safari may send the audio to Apple. Termivoa sends no audio anywhere. Without a browser speech service, Voice is hidden: use your keyboard’s microphone key.