phone-voice

What This Skill Does

The phone-voice skill transforms your OpenClaw agent into a sophisticated, phone-capable conversationalist. By integrating Twilio for telephony and ElevenLabs for low-latency voice synthesis, this skill allows users to interact with their agent via real-time voice calls. Unlike basic voice assistants, this bridge provides deep context awareness by injecting data from your local MEMORY.md and USER.md files, ensuring the agent remains aware of your historical preferences and ongoing tasks. The system includes built-in security features like caller ID authentication and voice PINs, along with proactive call screening to prevent unauthorized access. Furthermore, it manages costs and records high-quality transcripts, making it an essential tool for personal assistants, customer support automation, or voice-controlled home management.

Installation

1. Ensure your OpenClaw environment is active and running.
2. Execute clawhub install openclaw/skills/skills/cortexuvula/phone-voice to pull the necessary bridge dependencies.
3. Configure your .env file with your Anthropic, ElevenLabs, and Twilio API keys.
4. Define your caller whitelist in contacts.json to enable screening.
5. Deploy the FastAPI bridge using a secure tunnel. We recommend Cloudflare Tunnel for long-term stability, though ngrok is suitable for development. Point your ElevenLabs Custom LLM setting to your tunnel URL with the required Bearer authentication.

Use Cases

• Personal Productivity: Call your bot while driving to dictate notes, update your task list, or check your schedule.
• Home Automation: Control integrated OpenClaw devices through natural language voice commands.
• Customer Service: Provide a sophisticated, memory-aware phone interface for clients that references specific project documentation.
• Safety & Monitoring: Use the bot to screen incoming calls and summarize urgent messages before you answer.

Example Prompts

1. "Call my assistant and ask it to summarize the project notes from MEMORY.md for the meeting starting at 2 PM."
2. "Update my caller whitelist in the phone-voice settings to allow my project manager access to the voice agent."
3. "Check the cost tracking for my phone-voice usage over the last 30 days and provide a summary of total call time."

Tips & Limitations

• Latency: Ensure your server is hosted in a region close to your target audience to minimize voice delay.
• Security: Never expose your bridge directly to the open web without authentication. Always use the Bearer token feature.
• Memory: Keep your MEMORY.md file concise to avoid exceeding token limits during the initial conversation load.
• Cost: Be aware that ElevenLabs and Twilio charge per minute; monitor your usage logs frequently via the built-in cost tracking feature.