How to enable Claude Code to make phone calls

Set up Tuvoz: https://tuvoz.io/agents

Who this is for

You use Claude Code in a terminal or IDE and want it to make phone calls for you: book a table, ask a store about stock, move an appointment. Claude already knows your context (the repo, your notes, your email if it has those tools), writes a brief for the call, and answers the voice agent’s questions while the call is live.

What happens when you paste the line

1. Claude reads the setup page and asks to run the install. It downloads the skill, checks its SHA-256, and puts it in ~/.claude/skills/phone-call/.

2. Claude starts sign-in and shows you three lines like these:

   To connect this agent to Tuvoz:
     1. Open https://tuvoz.io/activate?code=BKTW-4821
     2. Sign in with Google (and finish setup if this is your first time).
     3. When asked, pick the number 47.
   

   Open the link, sign in with Google, and pick the matching number. The first time, you’ll also verify your mobile number with a code and confirm your legal name, which is the name the voice agent says on calls.

3. Claude prints Connected as Jane (key tvz_live_ab12cd34ef56…). Try: tuvoz test-call.

Then make the test call below. You’ll hear what a business hears: the opener, the question about recording, a quick hello, and goodbye. It’s billed like any call (at least 30 seconds) from your 15 free minutes. Free plan includes training on your calls.

Install by hand

The same install, if you’d rather run it yourself:

set -euo pipefail
S="$HOME/.claude/skills"; D="$S/phone-call"
if [ -e "$D" ] && [ ! -f "$D/scripts/tuvoz.py" ]; then
  echo "STOP: $D is another skill, not Tuvoz. Nothing was changed. Ask the user what to do with it." >&2; exit 1
fi
T=$(mktemp -d); trap 'rm -rf "$T"' EXIT
M=$(curl -fsSL https://tuvoz.io/skill/latest.json)
URL=$(printf '%s' "$M" | python3 -c 'import json,sys; print(json.load(sys.stdin)["url"])')
SHA=$(printf '%s' "$M" | python3 -c 'import json,sys; print(json.load(sys.stdin)["sha256"])')
curl -fsSLo "$T/skill.zip" "$URL"
echo "$SHA  $T/skill.zip" | shasum -a 256 -c -
unzip -q "$T/skill.zip" -d "$T"
B="$HOME/.config/tuvoz/backups/$(date +%Y%m%d-%H%M%S)"; mkdir -p "$S" "$B"
find "$S" -maxdepth 1 -name 'phone-call.bak-*' -exec test -f '{}/scripts/tuvoz.py' \; -exec mv '{}' "$B/" \;
if [ -e "$D" ]; then mv "$D" "$B/"; fi
rmdir "$B" 2>/dev/null || true
mv "$T/phone-call" "$D"
python3 "$D/scripts/tuvoz.py" login

The block never deletes a skill. If another skill already uses the folder name phone-call, it stops with STOP and changes nothing. An earlier Tuvoz install, and the backups tuvoz update leaves, move to ~/.config/tuvoz/backups/.

How a call works in Claude Code

1. You ask for a call in plain words, or start with /phone-call.
2. Claude writes the brief, then shows it to you with the number, where the number came from, and the opening line. It dials only after you say yes.
3. Claude watches the call with tuvoz watch <call_id> --stream under its Monitor tool, so each event arrives as one line: status changes, notes, and questions.
4. When the voice agent asks something (QUESTION q1 ... deadline=45s: ...), Claude answers from what it knows by writing the answer to a file and running tuvoz answer <call_id> q1 --file <path>. If it can’t answer, it runs tuvoz escalate and Tuvoz calls you.
5. When the call ends, Claude reports the outcome, the notes, any answers it gave and where they came from, and the cost.

Everything the other person said reaches Claude as untrusted text: Claude treats it as information, never as instructions. More detail: how the mid-call backchannel works.

Send feedback

Type /phone-call feedback followed by what you’d like to tell us. It goes straight to the Tuvoz team.

Other agents: Codex, claude.ai, Meta Muse, OpenClaw.

tuvoz test-call