Openclaw-config vapi-calls

Vapi Voice Calls

install

source · Clone the upstream repo

git clone https://github.com/TechNickAI/openclaw-config

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/TechNickAI/openclaw-config "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/vapi-calls" ~/.claude/skills/technickai-openclaw-config-vapi-calls && rm -rf "$T"

OpenClaw · Install into ~/.openclaw/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/TechNickAI/openclaw-config "$T" && mkdir -p ~/.openclaw/skills && cp -r "$T/skills/vapi-calls" ~/.openclaw/skills/technickai-openclaw-config-vapi-calls && rm -rf "$T"

manifest: skills/vapi-calls/SKILL.md

source content

Vapi Voice Calls

Make real outbound phone calls through Vapi's voice AI platform.

Prerequisites

Check that these are set in your environment:

```
VAPI_API_KEY
```
```
VAPI_ASSISTANT_ID
```
```
VAPI_PHONE_NUMBER_ID
```

If any are missing, read references/setup.md and complete first-time setup before proceeding.

Making a Call

Use the Vapi

POST /call

endpoint with:

```
phoneNumberId
```
and
```
assistantId
```
from your config
```
customer.number
```
— the phone number to call (E.164 format)
```
assistantOverrides.model.systemPrompt
```
— the full system prompt including task-specific instructions for this call
```
assistantOverrides.firstMessage
```
— a contextual opening line for this specific call

The base assistant defines voice, transcriber, and personality defaults. The override injects what THIS call is about.

System Prompt Structure for Each Call

The base prompt (heart-centered foundation + agent personality + "You are on a live phone call")
A
```
TASK:
```
section with specific instructions for this call

Everything the voice agent needs to know goes in the override. The voice agent has no memory between calls.

After the Call

Poll

GET /call/{id}

to check status and

endedReason

. Key values:

```
assistant-ended-call
```
— completed normally
```
customer-did-not-answer
```
— no pickup, consider retrying or texting first
```
pipeline-error-*
```
— voice provider issue, check configuration

Notes

Vapi built-in voices are free. ElevenLabs voices require a paid ElevenLabs plan.
Cost is ~$0.05-0.07/min, pay-as-you-go.
Template variables (
```
{{var}}
```
) can be used in prompts, resolved from call metadata.