A developer-centric toolkit for building and managing sophisticated Voice AI agents across Vapi, Bland.ai, and Retell platforms.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install voice-agent-pro
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install voice-agent-pro using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Voice Agent Builder?
Voice Agent Builder is a robust technical skill designed to bridge the gap between AI agents and real-world voice communication. By integrating with leading providers like Vapi, Bland.ai, and Retell, this skill enables developers to create, configure, and manage voice-first AI agents directly through the Openclaw Skills ecosystem. It handles the complexities of text-to-speech (TTS) synchronization, low-latency processing, and phone number provisioning, allowing for natural, human-like interactions in various business contexts.
Whether you are building a simple outbound notification system or a complex, multi-turn IVR, this skill provides the necessary abstractions to manage agent personalities, voice profiles from providers like ElevenLabs, and conversational logic. As part of the Openclaw Skills suite, it offers a unified interface to streamline voice operations without needing to write platform-specific boilerplate code for every provider.
Voice Agent Builder Use Cases
- Automating outbound lead qualification and sales calls.
- Creating intelligent voice assistants for inbound customer support and FAQ handling.
- Scheduling and managing appointments by linking voice agents to calendar APIs.
- Building custom IVR systems that use natural language instead of traditional DTMF menus.
- Integrating voice capabilities into CRM workflows for real-time contact updates and follow-ups.
How Voice Agent Builder Works
- Set up your environment variables with the necessary API keys for Vapi, Bland, or Retell to authenticate with Openclaw Skills.
- Configure your agent by defining the system prompt, selecting a voice provider like ElevenLabs, and setting conversational parameters.
- Provision a new phone number or import an existing one from Twilio or Vonage through the integrated management scripts.
- Initiate outbound calls via the CLI or configure the agent to handle inbound traffic directed to your provisioned numbers.
- Monitor call progress and retrieve full transcripts and metadata once the conversation concludes for further analysis.
Voice Agent Builder Setup
To deploy this skill within the Openclaw Skills framework, you must have Python 3.10+ and a Vapi API key.
First, set your environment variable:
export VAPI_API_KEY="your-vapi-api-key"
Create a new voice agent with a customized prompt:
python3 scripts/vapi_agent.py create-agent '{"name":"Support Agent","firstMessage":"Hello! How can I assist you?","systemPrompt":"You are a helpful support agent."}'
To make an outbound call, use the following command:
python3 scripts/vapi_agent.py call '{"assistantId":"YOUR_ASST_ID","phoneNumberId":"YOUR_PHONE_ID","customer":{"number":"+15551234567"}}'
Voice Agent Builder Data Schema & Taxonomy
The skill utilizes structured JSON to define agent behavior and call parameters. The primary configuration schema includes:
| Attribute | Type | Description |
|---|---|---|
name |
String | The internal identifier for the agent |
voice |
Object | Contains the provider (e.g., 11labs) and voiceId |
model |
Object | Specifies the LLM provider and model (e.g., gpt-4o) |
systemPrompt |
String | The core behavioral instructions for the AI |
serverUrl |
String | The webhook URL for receiving call events and data |
All call logs, including duration, cost, and transcriptions, are stored and accessible via the Openclaw Skills management interface.
Voice Agent Builder Advanced Features
- Support for multiple TTS engines including ElevenLabs, PlayHT, Deepgram, and Azure for varied voice quality and cost options.
- Built-in RAG capabilities allowing agents to query uploaded PDF or text documents during live conversations.
- Advanced function calling to trigger external API actions, such as database lookups or booking software, mid-call.
- Cross-platform comparison tools to evaluate latency, pricing, and features between Vapi, Bland.ai, and Retell.
- Webhook integration for real-time CRM updates, allowing Openclaw Skills to sync call outcomes directly with GoHighLevel or other platforms.
SKILL.md
Loading
Related Openclaw Skills
Voice Agent Builder
A comprehensive toolkit for building, configuring, and managing AI-powered voice agents across multiple telephony providers.
USAspending
A comprehensive CLI tool for retrieving federal spending data, contracts, grants, and agency information directly from USAspending.gov.
Uploadthing
A zero-dependency Python utility for AI agents to upload, manage, and deliver files via the Uploadthing API and CDN.
Toast Restaurant Management
A comprehensive integration for the Toast POS system allowing AI agents to manage restaurant orders, menus, and staff operations.
Voice Transcriber
A high-performance audio transcription and voice note archival tool powered by Deepgram Nova-3 for Openclaw Skills users.
Voice Transcriber
An AI-powered voice transcription and archival tool that converts audio files into searchable text for agent workflows.
Featured*
