voice-agent-builder
Build and manage Voice AI agents using Vapi, Bland.ai, or Retell. Create agents, configure voices, set prompts, make outbound calls, and retrieve transcripts. Includes platform comparison guide. Use when building phone agents, IVR systems, or voice-first customer service.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/aiwithabidi/voice-agent-builderVoice Agent Builder
Build, configure, and manage Voice AI agents. Supports Vapi (primary), Bland.ai, and Retell platforms.
Quick Start
export VAPI_API_KEY="your-vapi-api-key"
# Create a voice agent
python3 {baseDir}/scripts/vapi_agent.py create-agent '{"name":"Sales Agent","firstMessage":"Hi! How can I help you today?","systemPrompt":"You are a helpful sales assistant for Acme Corp."}'
# Make an outbound call
python3 {baseDir}/scripts/vapi_agent.py call '{"assistantId":"asst_xxx","phoneNumberId":"pn_xxx","customer":{"number":"+15551234567"}}'
# List agents
python3 {baseDir}/scripts/vapi_agent.py list-agents
# List calls
python3 {baseDir}/scripts/vapi_agent.py list-calls
Platform Comparison
| Feature | Vapi | Bland.ai | Retell |
|---|---|---|---|
| Best For | Custom agents, dev-friendly | Simple outbound campaigns | Enterprise, low latency |
| Latency | ~800ms | ~500ms | ~500ms |
| Languages | 100+ | 30+ | 30+ |
| Custom LLM | ✅ Any OpenAI-compatible | ✅ Limited | ✅ Via API |
| Phone Numbers | Buy/import | Buy/import | Buy/import |
| Pricing | $0.05/min + provider costs | $0.09/min all-in | $0.07-0.15/min |
| WebSocket | ✅ | ❌ | ✅ |
| Knowledge Base | ✅ Built-in | ✅ | ✅ |
| Transfers | ✅ | ✅ | ✅ |
Recommendation: Start with Vapi — most flexible, best docs, largest community. Use Bland for simple high-volume outbound. Use Retell for enterprise low-latency needs.
See {baseDir}/scripts/voice_comparison.md for detailed breakdown.
Agent Creation Workflow
1. Choose a Voice
Vapi supports multiple TTS providers:
- ElevenLabs — Best quality, most natural (recommended)
- PlayHT — Good quality, lower cost
- Deepgram — Fast, good for real-time
- Azure — Enterprise, many languages
2. Configure the Agent
{
"name": "Appointment Setter",
"firstMessage": "Hi! This is Sarah from Dr. Smith's office. I'm calling to help you schedule your appointment.",
"systemPrompt": "You are Sarah, a friendly appointment scheduler...",
"voice": {
"provider": "11labs",
"voiceId": "21m00Tcm4TlvDq8ikWAM"
},
"model": {
"provider": "openai",
"model": "gpt-4o",
"temperature": 0.7
},
"endCallFunctionEnabled": true,
"maxDurationSeconds": 300,
"silenceTimeoutSeconds": 30
}
3. Prompt Engineering for Voice
Voice prompts differ from text. Key principles:
- Keep responses SHORT — 1-2 sentences max per turn
- Be conversational — use filler words naturally ("Sure thing!", "Got it!")
- Handle interruptions — voice agents get cut off, design for it
- Confirm understanding — repeat back key info (names, numbers, dates)
- Include fallback — "I didn't catch that, could you repeat?"
4. Phone Number Setup
# List available phone numbers
python3 {baseDir}/scripts/vapi_agent.py list-phones
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-aiwithabidi-voice-agent-builder": {
"enabled": true,
"auto_update": true
}
}
}Related Skills
freshsales
Freshsales CRM integration — manage contacts, leads, deals, accounts, tasks, and sales sequences via the Freshsales API. Track deal pipelines, automate lead assignments, log activities, and generate sales reports. Built for AI agents — Python stdlib only, no dependencies. Use for sales CRM, contact management, deal tracking, pipeline reporting, and sales automation.
gemini-video-analyzer
Native video analysis using Google Gemini API. Upload and analyze video files — describe scenes, extract text/UI, answer questions about content, transcribe speech, identify objects and actions. Use when: (1) User sends a video file and wants it analyzed, (2) Video summarization or description needed, (3) Extracting text, UI elements, or information from screen recordings, (4) Answering questions about video content, (5) Comparing multiple videos, (6) Analyzing tutorials, demos, or walkthroughs.
agent-memory
Full AI agent memory stack — Mem0 unified memory engine with vector search (Qdrant) and knowledge graph (Neo4j), plus SQLite for structured data. Complete setup script and tools. Give your OpenClaw agent a real brain with semantic recall, entity relationships, and structured storage.
neon
Neon serverless Postgres — manage projects, branches, databases, roles, endpoints, and compute via the Neon API. Create database branches for development, manage connection endpoints, scale compute, and monitor usage. Built for AI agents — Python stdlib only, zero dependencies. Use for serverless Postgres, database branching, database management, development workflows, and cloud database automation.
onepassword
1Password Connect — vaults, items, secrets management for server-side applications.