Use the Speak AI MCP server tool create_voice_agent_from_prompt to create a new
voice agent from a plain-English description, via an LLM call.
What does it do?
Create a new voice agent from a plain-English description, via an LLM call. Requires
the OWNER or ADMIN role. Saves a draft agent, then generates and saves its name,
personality, instructions, chat settings, and default voice, speech-to-text, and LLM
settings. Does not assign a phone number. The response includes the new agentId plus
either the generated agent, or needsFollowUp: true with a followUpQuestion when the
prompt is too vague. In that case the draft agent still exists with placeholder
settings; call generate_voice_agent_config on that agentId with more detail.
Parameters
create_voice_agent_from_prompt takes 3 parameters, 1 of them required.
| Name | Type | Required | Description |
|---|---|---|---|
prompt |
string | Yes | Plain-English description of the agent to build, e.g. “a friendly dental clinic receptionist that books appointments and answers insurance questions”. |
name |
string | No | Initial name for the draft agent. Replaced by the generated name when generation succeeds; kept only if the response asks a follow-up question. |
manualInstructions |
string | No | Requirements the generated instructions must include. Generation still runs; when this is sent, no follow-up question is returned even if the prompt is vague. |
Behaviour
The Speak AI MCP server publishes these behaviour hints for
create_voice_agent_from_prompt.
- Can change data in your workspace.
- Does not delete or overwrite existing data.
- Repeating an identical call can have an extra effect.
- Stays inside Speak AI. Does not reach other systems.
Related
- Voice agents lists the rest of the Voice agents category.
- Tool reference lists every Speak AI MCP server tool.
- MCP server covers setup, supported clients, and what agents can do once connected.