Dial MCP Server
Enables AI agents to provision phone numbers, send SMS, place AI voice calls, and react to inbound events via the Dial communication stack, all through MCP tools.
README
@getdial/cli
The official command-line interface for Dial — a communication stack for AI agents. Provision phone numbers, send SMS, place AI voice calls, and react to inbound events, all from your terminal. The CLI wraps the Dial REST API so you never have to write HTTP code.
Installation
Install globally from npm:
npm install -g @getdial/cli
Or use the bootstrap script, which installs the CLI and walks you through sign-up:
curl -fsSL https://getdial.ai/install | bash
Requires Node.js 22.19+.
Quick start
dial signup you@example.com # email a 6-digit sign-up code
dial onboard --code 123456 \ # verify the code and provision your account
--inbound-instruction "You are my receptionist. Greet the caller and find out what they need."
dial doctor # check account state and what to do next
Once onboarded, your API key is saved locally and the CLI uses it automatically.
# Send an SMS
dial message --to +14155550123 --body "Hello from Dial"
# Place an AI voice call
dial call --to +14155550123 --outbound-instruction "You are a helpful scheduling assistant."
# Wait for an inbound event (e.g. a verification code)
dial wait-for message.received --field to=+14155550123
Commands
| Command | Description |
|---|---|
dial doctor |
Report account state and what to do next. |
dial signup <email> |
Email a 6-digit sign-up code. |
dial onboard --code <code> |
Verify the code and finish onboarding. |
dial number list |
List the phone numbers on your account. |
dial number purchase |
Purchase an additional phone number. |
dial number set <number> |
Update a number's inbound instruction. |
dial message |
Send an SMS. |
dial message list |
List recent messages. |
dial message reply |
Reply or react to a message. |
dial call |
Place an outbound AI voice call. |
dial call list |
List recent calls. |
dial call get <id> |
Fetch a single call — status, duration, transcript. |
dial wait-for <event> |
Block until a matching account event arrives. |
dial listen install |
Install the background event daemon. |
dial listen status |
Report the daemon's state and recent events. |
dial listen uninstall |
Stop and remove the daemon. |
dial local-target add url <url> |
Fan out events to a local HTTP endpoint. |
dial local-target add cmd <path> |
Run an executable once per event. |
dial local-target list |
List registered fan-out targets. |
dial local-target remove <id> |
Unregister a fan-out target. |
dial mcp |
Run a local stdio MCP server exposing every command as an agent tool. |
Run dial --help for the full command tree, or dial <command> --help for a specific command's flags. Every command accepts --json for machine-readable output (except dial mcp, which speaks JSON-RPC on stdout).
Configuration
| Variable | Description |
|---|---|
DIAL_API_URL |
Target a non-default Dial deployment. |
Your API key is stored at ~/.local/share/dial/auth.json (honoring XDG_DATA_HOME).
Use from an AI agent
The CLI ships with a skill that teaches AI coding agents how to drive dial. Install it into your agent's config during onboarding:
dial onboard --code 123456 --agent claude-code
Supported agents: claude-code, cursor, codex, opencode, pi, openclaw, nanoclaw, hermes.
Local MCP server
dial mcp runs a local Model Context Protocol server over stdio, exposing every command as an agent tool. Point a local MCP client at dial mcp as the server command — it reuses the API key saved by dial onboard (no OAuth, no config). It's the local counterpart to the hosted Remote MCP server, with the same operational tools plus the local-only verbs (signup, onboard, listen, local-target).
# Claude Code, for example
claude mcp add dial -- dial mcp
See the Local MCP docs for client setup.
Documentation
Full documentation lives at docs.getdial.ai — including the CLI reference and the listen service guide.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。