SC

skeptrunedev/call4me

Developer tools

このページは AI により自動翻訳されています。技術的な詳細は原文をご確認ください。 原文を見る

16 stars 1 forks 品質 83 トレンド 30MIT

Your coding agent makes phone calls for you: an MCP tool on Telnyx + OpenAI GPT-Live

概要

call4me lets your AI agent make real phone calls for you. Ask Claude, ChatGPT, Cursor, Codex or your own app to book a table, schedule a dentist appointment, cancel a subscription or ask a business a question, and it calls, waits on hold, gets through the phone menu, talks to the person who answers and reports back with the outcome and a transcript. It can also ring you to join the call. Calls are paid from prepaid credits, starting at $10. The hosted MCP server is https://call4.me/mcp (streamable HTTP). Your agent signs you in through the browser with Google or X. The easiest setup is to paste the prompt from call4.me/mcp into your agent. To add it by hand: Cursor: add it to ~/.cursor/mcp.json and sign in when Cursor asks. In your own code, the Vercel AI SDK package gives an agent the same tools: - Examples: real calls and the prompts that placed them. - Blog: guides for Claude Code, the AI SDK and more. - MCP Registry: io.github.skeptrunedev/call4me.

README

以下は原文 README の抜粋です。原文を見る。

call4me: MCP server for making phone calls

call4me is a remote MCP server for making phone calls. Point any MCP client at https://call4.me/mcp (streamable HTTP) and your agent can phone a business for you. One-line install for Claude Code:

claude mcp add --scope user --transport http call4me https://call4.me/mcp

https://call4.me

call4me lets your AI agent make real phone calls for you. Ask Claude, ChatGPT, Cursor, Codex or your own app to book a table, schedule a dentist appointment, cancel a subscription or ask a business a question, and it calls, waits on hold, gets through the phone menu, talks to the person who answers and reports back with the outcome and a transcript. It can also ring you to join the call. Calls are paid from prepaid credits, starting at $10.

Use it

The hosted MCP server is https://call4.me/mcp (streamable HTTP). Your agent signs you in through the browser with Google or X. The easiest setup is to paste the prompt from call4.me/mcp into your agent. To add it by hand:

claude code:     claude mcp add --scope user --transport http call4me https://call4.me/mcp   then /mcp, call4me, authenticate
codex:           codex mcp add call4me --url https://call4.me/mcp   then   codex mcp login call4me
claude desktop / claude.ai / chatgpt:  settings, connectors, add custom connector: https://call4.me/mcp
can't sign in?   create a key at https://call4.me/account and use https://call4.me/mcp/, or send "Authorization: Bearer " to /mcp

Cursor: add it to ~/.cursor/mcp.json and sign in when Cursor asks.

{
  "mcpServers": {
    "call4me": { "url": "https://call4.me/mcp" }
  }
}

In your own code, the Vercel AI SDK package gives an agent the same tools:

npm install call4me-ai-sdk ai zod @ai-sdk/openai
import { call4meTools } from 'call4me-ai-sdk';
// with CALL4ME_API_KEY set (from https://call4.me/account):
// generateText({ model, prompt: 'Call ... and ask ...', tools: call4meTools() })

The rest of this README is for working on call4me itself or hosting your own copy.

How a call works

agent ──MCP──▶ worker ──POST /v2/calls──▶ Telnyx ──PSTN──▶ business
                  ▲                          │ media stream (PCMU, bidirectional RTP)
                  │ webhooks                 ▼
GitHub で README 全文を見る

インストール

This server does not publish a one-line install command.

Open the repository installation guide MCP 初心者?mcpskill.org のガイドから始めましょう →

設定

{ "mcpServers": { "call4me": { "url": "https://call4.me/mcp" } } }