Prerequisites
- A Vapi account with at least one configured assistant
- A Vapi private API key (Bearer token)
- project_contributor access in AgentRuntime
Connect in AgentRuntime
Get a Vapi API key
Sign in to the Vapi dashboard and copy your private API key from settings.
Create a connection
Go to Connections and click New custom connection to create an API key connection with
APIKey.Add vapi MCP instance
Go to MCP, click Add instance for vapi, wire your connection on Instance config, set the profile active, and save.
This adapter currently exposes read-only assistant discovery tools. Outbound call creation and live call control are configured in Vapi directly or via future adapter tools.
What you can build
- Assistant inventory — Scheduled
list_assistants→ diff against expected list → alert on drift. - Config documentation —
get_assistant→ LLM summarizes model, voice, and tool settings → save to Notion. - Pre-call validation — Before a campaign, verify assistant ID and prompt version match release notes.
- Cross-system sync — Export
raw_jsonto internal config store for compliance review.
Tools
| Tool | Description |
|---|---|
list_assistants | List voice assistants visible to the API key |
get_assistant | Full assistant configuration by ID (includes pretty-printed JSON) |
Example
Fetch assistant configuration for audit logging:Configuration
| Key | Required | Description |
|---|---|---|
APIKey | Yes | Vapi private API key (Bearer token) |
APIBase | No | Vapi API base URL |
Troubleshooting
| Issue | Fix |
|---|---|
401 Unauthorized | Verify private key, not a publishable or webhook-only key |
| Assistant not in list | Key may be scoped to a different Vapi org or project |
Empty raw_json | Check HTTP status in tool output; assistant ID may be invalid |
| Timeout on get | Large assistant configs with many tools may need higher timeout_s |
Related
- Integrations quickstart
- ElevenLabs connector — text-to-speech synthesis
- Connector catalog
- Troubleshooting