---

name: ops-voice description: Voice operations — make phone calls (Bland AI), text-to-speech (ElevenLabs), transcribe audio (Whisper/Groq). Replace OpenClaw voice capabilities. argument-hint: "[call|tts|transcribe|setup]" allowed-tools:

• Bash
• Read
• Write
• AskUserQuestion
• WebFetch

---

OPS:VOICE — Voice Operations

Voice interface commands. All API calls via curl — no SDK dependencies.

Credential resolution order: userConfig → env vars → Doppler MCP tools (mcp__doppler__*) → Doppler CLI fallback (doppler secrets get <KEY> --plain) → password manager

---

Sub-commands

Parse $ARGUMENTS for the command keyword, then execute:

---

call [phone] [prompt] — Bland AI phone call

Requires: bland_ai_api_key in userConfig or BLAND_AI_API_KEY env or Doppler.

BLAND_KEY="${BLAND_AI_API_KEY:-$(doppler secrets get BLAND_AI_API_KEY --plain 2>/dev/null || true)}"
PHONE="<extracted from $ARGUMENTS>"
PROMPT="<extracted from $ARGUMENTS or ask user>"
MAX_DURATION="${BLAND_MAX_DURATION:-300}"  # seconds
VOICE="${BLAND_VOICE:-male}"

# Make the call
RESPONSE=$(curl -s -X POST "https://api.bland.ai/v1/calls" \
  -H "authorization: $BLAND_KEY" \
  -H "Content-Type: application/json" \
  -d "{
    \"phone_number\": \"$PHONE\",
    \"task\": \"$PROMPT\",
    \"voice\": \"$VOICE\",
    \"max_duration\": $MAX_DURATION,
    \"record\": true
  }")

CALL_ID=$(echo "$RESPONSE" | python3 -c "import json,sys; print(json.load(sys.stdin).get('call_id',''))" 2>/dev/null)

# Poll for completion (up to 5 min)
if [ -n "$CALL_ID" ]; then
  echo "Call initiated: $CALL_ID"
  for i in $(seq 1 30); do
    sleep 10
    STATUS=$(curl -s "https://api.bland.ai/v1/calls/$CALL_ID" \
      -H "authorization: $BLAND_KEY" | \
      python3 -c "import json,sys; d=json.load(sys.stdin); print(d.get('status',''), d.get('transcripts','')[-1].get('text','') if d.get('transcripts') else '')" 2>/dev/null)
    echo "Status: $STATUS"
    [[ "$STATUS" == completed* ]] && break
  done
fi

Output: Call ID, live status, transcript when complete.

---

tts [text] [--voice voice_id] [--out file.mp3] — ElevenLabs text-to-speech

Requires: elevenlabs_api_key in userConfig or ELEVENLABS_API_KEY env or Doppler.

EL_KEY="${ELEVENLABS_API_KEY:-$(doppler secrets get ELEVENLABS_API_KEY --plain 2>/dev/null || true)}"
VOICE_ID="${ELEVENLABS_VOICE_ID:-21m00Tcm4TlvDq8ikWAM}"  # Rachel (default)
TEXT="<extracted from $ARGUMENTS>"
OUT_FILE="${OUT_FILE:-/tmp/ops-tts-$(date +%s).mp3}"

# List voices if voice name provided (not an ID)
# Synthesize
curl -s -X POST "https://api.elevenlabs.io/v1/text-to-speech/${VOICE_ID}" \
  -H "xi-api-key: $EL_KEY" \
  -H "Content-Type: application/json" \
  -d "{
    \"text\": \"$TEXT\",
    \"model_id\": \"eleven_monolingual_v1\",
    \"voice_settings\": {\"stability\": 0.5, \"similarity_boost\": 0.75}
  }" \
  --output "$OUT_FILE"

echo "Audio saved to: $OUT_FILE"
# Auto-play on macOS
command -v afplay &>/dev/null && afplay "$OUT_FILE" &

Output: Audio file path. Auto-plays on macOS via afplay.

---

transcribe [file_path] — Groq Whisper transcription

Requires: groq_api_key in userConfig or GROQ_API_KEY env or Doppler.

GROQ_KEY="${GROQ_API_KEY:-$(doppler secrets get GROQ_API_KEY --plain 2>/dev/null || true)}"
AUDIO_FILE="<extracted from $ARGUMENTS>"

if [ ! -f "$AUDIO_FILE" ]; then
  echo "ERROR: File not found: $AUDIO_FILE"
  exit 1
fi

TRANSCRIPT=$(curl -s -X POST "https://api.groq.com/openai/v1/audio/transcriptions" \
  -H "Authorization: Bearer $GROQ_KEY" \
  -F "file=@$AUDIO_FILE" \
  -F "model=whisper-large-v3" \
  -F "response_format=json" | \
  python3 -c "import json,sys; print(json.load(sys.stdin).get('text',''))" 2>/dev/null)

echo "$TRANSCRIPT"

Output: Transcript text printed to stdout.

---

setup — Configure voice API keys

Before asking for anything, auto-scan ALL sources in a single background batch:

# Env vars
printenv BLAND_AI_API_KEY BLAND_API_KEY ELEVENLABS_API_KEY GROQ_API_KEY 2>/dev/null

# Shell profiles
grep -h 'BLAND\|ELEVENLABS\|GROQ' ~/.zshrc ~/.bashrc ~/.zprofile ~/.envrc 2>/dev/null | grep -v '^#'

# Doppler — ALL projects
for proj in $(doppler projects --json 2>/dev/null | jq -r '.[].slug'); do
  for cfg in dev stg prd; do
    doppler secrets --project "$proj" --config "$cfg" --json 2>/dev/null | \
      jq -r --arg proj "$proj" --arg cfg "$cfg" 'to_entries[] | select(.key | test("BLAND|ELEVENLABS|GROQ"; "i")) | "\(.key)=\(.value.computed | .[0:12])... (doppler:\($proj)/\($cfg))"'
  done
done

# Dashlane
dcli password bland --output json 2>/dev/null | jq -r '.[] | select(.password != null) | "\(.title): key found"'
dcli password elevenlabs --output json 2>/dev/null | jq -r '.[] | select(.password != null) | "\(.title): key found"'
dcli password groq --output json 2>/dev/null | jq -r '.[] | select(.password != null) | "\(.title): key found"'

# Keychain
security find-generic-password -s "bland-ai-api-key" -w 2>/dev/null
security find-generic-password -s "elevenlabs-api-key" -w 2>/dev/null
security find-generic-password -s "groq-api-key" -w 2>/dev/null

Present all findings. Only prompt for keys NOT found in any source. Then validate each found key in background:

1. Bland AI: curl -s -H "authorization: $KEY" https://api.bland.ai/v1/me — check balance
2. ElevenLabs: curl -s -H "xi-api-key: $KEY" https://api.elevenlabs.io/v1/voices?page_size=1 — list voices
3. Groq: curl -s -H "Authorization: Bearer $KEY" https://api.groq.com/openai/v1/models — list models

Report: [service] ✓ connected or [service] ✗ invalid key — [error]

---

Execution

1. Resolve the sub-command from $ARGUMENTS (first word: call / tts / transcribe / setup)
2. Resolve credentials in order: env → Doppler
3. Execute the matching curl block above
4. If a required key is missing and setup was not invoked, suggest /ops:ops-voice setup