Claude Code Voice
Voice conversations with Claude Opus 4.5 about your code.
Talk through problems, brainstorm ideas, or get a code review — all over the phone.
Quick Start
# Install
pip install claude-code-voice
# One-time setup
claude-code-voice setup
# Register a project
cd your-project
claude-code-voice register
# Start everything (server + tunnel + config)
claude-code-voice start
That's it! Now you can:
- Have Claude call you:
claude-code-voice call "debug the auth flow" - Call Claude: Dial your Vapi number and Claude answers with your project loaded
What You Need
Before running setup:
- Vapi account — https://vapi.ai (free to sign up, $10 credit included)
- Vapi API key — https://dashboard.vapi.ai/api-keys
- Vapi phone number — https://dashboard.vapi.ai/phone-numbers (~$2/month)
- Node.js — for localtunnel
Note: New Vapi accounts get $10 free credit — enough to test everything without spending anything!
Features
| Feature | Description |
|---|---|
| Opus 4.5 | Best-in-class reasoning for technical discussions |
| Project context | Git status, recent files, todos loaded automatically |
| Live tools | Claude reads files and searches code during calls |
| Auto-transcripts | Every call saved as markdown |
| Personalized | Claude greets you by name |
| Inbound calls | Call your number and Claude answers |
Commands
claude-code-voice setup # Configure API key, phone, name
claude-code-voice register # Register current project
claude-code-voice start # Start server + tunnel (recommended)
claude-code-voice call [topic] # Have Claude call you
claude-code-voice status # Check configuration
claude-code-voice config name <name> # Update your name
claude-code-voice config show # Show all config
claude-code-voice history # View past calls
Customization
Change voice, model, or language:
# Voice (provider:voiceId)
claude-code-voice config voice openai:nova # OpenAI Nova
claude-code-voice config voice elevenlabs:rachel # ElevenLabs Rachel
claude-code-voice config voice deepgram:asteria # Deepgram Asteria
# Model
claude-code-voice config model claude-sonnet-4-20250514 # Use Sonnet 4
# Language
claude-code-voice config language es # Spanish
claude-code-voice config language fr # French
Supported voice providers: openai, elevenlabs, deepgram, playht, azure
How It Works
You ──call──▶ Vapi Phone ──webhook──▶ Your Server ──context──▶ Claude Opus 4.5
│
reads your code
- Setup stores your Vapi credentials and creates tools
- Register snapshots project context (git, files, todos)
- Start runs server + tunnel, auto-configures Vapi
- Call — Claude has full context about your project
The start Command
The start command is the easiest way to get everything running:
claude-code-voice start
It automatically:
- Kills any existing server on port 8765
- Starts the context server
- Starts localtunnel and waits for URL
- Updates all Vapi tools with the new URL
- Configures inbound call webhooks
- Shows your Vapi number for inbound calls
Just keep that terminal open and you're ready for calls.
Transcripts
Transcripts auto-save to ~/.claude/skills/call/data/transcripts/ when calls end.
Each transcript includes:
- Full conversation
- AI-generated summary
- Call metadata (duration, project, topic)
Troubleshooting
"Connection issues" or "404 errors"
Run claude-code-voice start — it auto-configures everything when the tunnel URL changes.
"I don't recognize this number"
Call from the phone number you used during setup.
Claude can't read files during call
Make sure claude-code-voice start is running in a terminal.
Check your configuration
claude-code-voice status
As a Claude Code Skill
For /call directly in Claude Code:
git clone https://github.com/abracadabra50/claude-code-voice-skill.git
ln -s /path/to/claude-code-voice-skill ~/.claude/skills/call
Then use /call in conversations.
Manual Setup (Advanced)
If you prefer manual control over the server and tunnel:
# Terminal 1: Start server
claude-code-voice server
# Terminal 2: Start tunnel
npx localtunnel --port 8765
# Terminal 3: Configure (run after each tunnel restart)
claude-code-voice config server-url https://xxx.loca.lt
claude-code-voice configure-inbound
License
MIT
Metadata
Release files for claude-code-voice 0.1.5
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| claude_code_voice-0.1.5.tar.gz | 19.5 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| claude_code_voice-0.1.5-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 40.2 kB
Release files / claude_code_voice-0.1.5.tar.gz
| Download URL | claude_code_voice-0.1.5.tar.gz |
|---|---|
| Size | 19.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
ef06fb26043e67f370508d79836eb628992877bb7ab7ca47eb06b9062a23f5ed
|
|
BLAKE2b-256 checksum How to use checksums |
b1347ec4f025e1878ca81bb0252e455bbe4e34beca1e9fe3a11cd6f7af1eb10f
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jan 6, 2026.
Transparency logRelease files / claude_code_voice-0.1.5-py3-none-any.whl
| Download URL | claude_code_voice-0.1.5-py3-none-any.whl |
|---|---|
| Size | 20.7 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
22b07500f9c0c875a0a6395137d6aba6245eb3159ee45108e43cf5e2557ca374
|
|
BLAKE2b-256 checksum How to use checksums |
7481deeac50175d82f07bd955baff91dba87c695cd5bfdae014e71d817fff9b7
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/6.1.0 CPython/3.13.7
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Jan 6, 2026.
Transparency log