Retell AI — Claude Code MCP server: calls and agents
Wires Retell AI voice agents into your assistant: place a call, read call history, edit an agent prompt.
claude mcp add --env RETELL_API_KEY=<RETELL_API_KEY> --env MCP_WORKSPACE_PATH=<MCP_WORKSPACE_PATH> --transport stdio retell-ai -- npx -y @mindstone/mcp-server-retell-aiWhat Retell AI does
Retell AI is a local stdio MCP server that exposes Retell AI phone operations to your assistant. It ships 32 tools covering calls, agents, LLM prompts, voices, knowledge bases and chat history. A typical run inspects the agent and its prompt, picks a number, then dials.
Outbound calling is a real-world action. The host has to show you from_number and to_number and wait for your confirmation before every create_phone_call. Prior approvals do not carry over — each call gets confirmed on its own. Numbers are checked locally against E.164, so a malformed one comes back as INVALID_PHONE_NUMBER and never reaches Retell. Billing runs per minute on your Retell AI plan, which makes a stray call an actual expense.
create_batch_call fires one billed call per task, and the confirmation rules are the same: surface the whole recipient list first. Check get_concurrency before a campaign, or it can crowd out inbound capacity. Knowledge-base uploads read only inside MCP_WORKSPACE_PATH, falling back to the system temp directory when that is unset; .. traversal and escaping symlinks are refused before any disk read. You need Node.js 20+, npm and a RETELL_API_KEY.
What it helps with
- call a customer with a voice agent and keep the recording
- pull the transcript and status of a finished call
- edit an agent prompt before a campaign goes out
- check free capacity ahead of a batch dialer run
- upload documents into an agent knowledge base
Example prompts for the agent
- Call Jane with the support agent and confirm tomorrow's appointment
- Show me the status and transcript of the last call
- Check the support agent's prompt and its dynamic variables
How to install Retell AI
Platform CLI
claude mcp add --env RETELL_API_KEY=<RETELL_API_KEY> --env MCP_WORKSPACE_PATH=<MCP_WORKSPACE_PATH> --transport stdio retell-ai -- npx -y @mindstone/mcp-server-retell-aiReplace <…> with your own values
Platform CLI
codex mcp add retell-ai --env RETELL_API_KEY=<RETELL_API_KEY> --env MCP_WORKSPACE_PATH=<MCP_WORKSPACE_PATH> -- npx -y @mindstone/mcp-server-retell-aiReplace <…> with your own values
Config file
{
"mcpServers": {
"retell-ai": {
"command": "npx",
"args": [
"-y",
"@mindstone/mcp-server-retell-ai"
],
"env": {
"RETELL_API_KEY": "<RETELL_API_KEY>",
"MCP_WORKSPACE_PATH": "<MCP_WORKSPACE_PATH>"
}
}
}
}Replace <…> with your own values
Config file
{
"servers": {
"retell-ai": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@mindstone/mcp-server-retell-ai"
],
"env": {
"RETELL_API_KEY": "<RETELL_API_KEY>",
"MCP_WORKSPACE_PATH": "<MCP_WORKSPACE_PATH>"
}
}
}
}Replace <…> with your own values
Platform CLI
gemini mcp add -e RETELL_API_KEY=<RETELL_API_KEY> -e MCP_WORKSPACE_PATH=<MCP_WORKSPACE_PATH> retell-ai npx -y @mindstone/mcp-server-retell-aiReplace <…> with your own values
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Retell AI
How do I install Retell AI in Claude Code?
Run claude mcp add mcp-server-retell-ai -e RETELL_API_KEY=<RETELL_API_KEY> -- npx -y @mindstone/mcp-server-retell-ai. Swap in your own key from the Retell AI dashboard.
What does Retell AI do?
It hands your assistant Retell AI tools for calls, batch campaigns, agent setup, LLM prompts, voices, knowledge bases and chat history.
What does the server need to run?
Node.js 20+, npm and a RETELL_API_KEY. Set MCP_WORKSPACE_PATH if you want to upload knowledge-base files; otherwise reads stay inside the system temp directory.
