supercall_skill

This skill makes AI-powered phone calls with custom personas and ultra-low latency, enabling autonomous conversations, appointment confirmations, and message
  • Python

2.5k

GitHub Stars

7

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill supercall

  • _meta.json624 B
  • CHANGELOG.md2.9 KB
  • index.ts13.0 KB
  • openclaw.plugin.json2.6 KB
  • package.json1014 B
  • README.md3.9 KB
  • SKILL.md8.8 KB

Overview

This skill makes AI-powered phone calls using OpenAI Realtime API and Twilio, enabling ultra-low latency voice conversations with customizable personas and goals. It runs autonomous persona calls, streams bidirectional audio, and isolates the callee from the gateway agent to reduce attack surface. It supports a mock provider for testing and real Twilio integration for production-grade calls.

How this skill works

You configure provider credentials (OpenAI and Twilio) and a caller ID, then invoke actions such as persona_call to start an autonomous call that follows a specified persona, goal, and opening line. The skill opens a local webhook server and a WebSocket media stream to handle audio in realtime, forwards audio to the OpenAI Realtime API, and uses Twilio for PSTN connectivity. Call transcripts and status are persisted to disk and callbacks notify the host via a secure hooks token.

When to use it

  • Confirm appointments or availability without manual dialing
  • Deliver time-sensitive or repeated messages autonomously
  • Have an AI handle outbound phone conversations with a scripted persona
  • Test voice workflows locally with the mock provider before production
  • Streamline customer outreach where low-latency, natural conversation matters

Best practices

  • Protect and rotate OPENAI_API_KEY, TWILIO_ACCOUNT_SID, TWILIO_AUTH_TOKEN, and hooks.token regularly
  • Use a paid/custom tunnel domain (ngrok.app or Tailscale) in production to ensure strict webhook verification
  • Verify publicUrl or tunnel configuration before startup to avoid misdirected self-test requests
  • Review and purge transcripts in ~/clawd/supercall-logs to limit sensitive data at rest
  • Test with the mock provider first to validate persona behavior and callbacks

Example use cases

  • An assistant persona calls clients to confirm dinner or meeting availability and records confirmations
  • A logistics bot calls drivers to relay route updates and collects verbal acknowledgments
  • Automated appointment reminders with follow-up questions handled by a friendly persona
  • Quality assurance testing of call flows using the mock provider and scripted opening lines
  • Outbound survey calls where AI adapts follow-up questions based on responses

FAQ

You must provide an OpenAI API key and Twilio account credentials; an ngrok auth token is optional for tunnel setup.

Where are call transcripts stored?

Transcripts are saved to ~/clawd/supercall-logs; you should manage and rotate those logs to protect sensitive information.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational