SendSoon AI MCP
Enables AI agents to send emails, perform IP lookups, and convert files to Markdown.
README
SendSoon AI MCP
Use SendSoon email, IP lookup, and file-to-Markdown capabilities from Codex, Claude, Cursor, and other AI agents.
Configure it once, then describe tasks in natural language. You usually do not need to mention SendSoon in every prompt.
What you can do
| Tool | Purpose | Example |
|---|---|---|
send_email |
Send a single email | “Send me a test email” |
ip_lookup |
Look up public IP information | “Look up the location of 8.8.8.8” |
markitdown_convert |
Convert a file to Markdown | “Convert this PDF to Markdown” |
Try it in Google Colab
Skip local setup and try ip_lookup, markitdown_convert, and send_email in the browser.
The notebook calls the same SendSoon HTTP APIs used by these MCP tools. To use them from Cursor, Claude, or Codex, continue with the local MCP setup below.
Install MCP
Choose npm (Node) or PyPI (Python). Tool names and environment variables are the same.
npm (Node)
Published as @sendsoon/ai. Clients can start it with npx (downloads on first use):
npx -y @sendsoon/ai
Optional global install:
npm install -g @sendsoon/ai
sendsoon-mcp
Requires Node.js 20+.
PyPI (Python)
Published as sendsoon-mcp. Prefer uvx so no global install is needed:
uvx sendsoon-mcp
Or:
pip install sendsoon-mcp
sendsoon-mcp
Requires Python 3.10+. See python/README.md for details.
Prepare the configuration
All clients use the same environment variables:
| Setting | Required | What to enter |
|---|---|---|
SENDSOON_EMAIL_RECIPIENT |
Yes | Replace <YOUR_EMAIL> with your recipient address |
SENDSOON_API_KEY |
No | Leave empty for an unregistered trial. One public IP can send up to three free test emails per day. Enter your generated Key for continued use |
SENDSOON_API_BASE_URL |
No | Defaults to https://www.sendsoonai.com. Set it only to target a different environment |
Never commit a real Key to Git or share it with anyone.
Get an API Key
- Sign up or sign in on the SendSoon registration page.
- Open Profile and generate a Key in the API Key section.
- Copy the one-time
ssk_live_...Key immediately and enter it asSENDSOON_API_KEYin your MCP configuration. - Save the configuration and restart your MCP client. Requests with a valid Key do not use the anonymous IP daily quota.
If the anonymous quota is exhausted, send_email will tell you to register and configure a Key. An invalid or revoked Key is rejected and does not fall back to the anonymous quota.
Choose your client
Cursor
Open Settings > Tools & MCP in Cursor and add an MCP server. You can also save the following configuration as .cursor/mcp.json in your project or ~/.cursor/mcp.json in your user directory.
Node (npm):
{
"mcpServers": {
"sendsoon": {
"command": "npx",
"args": ["-y", "@sendsoon/ai"],
"env": {
"SENDSOON_EMAIL_RECIPIENT": "<YOUR_EMAIL>",
"SENDSOON_API_KEY": ""
}
}
}
}
Python (PyPI):
{
"mcpServers": {
"sendsoon": {
"command": "uvx",
"args": ["sendsoon-mcp"],
"env": {
"SENDSOON_EMAIL_RECIPIENT": "<YOUR_EMAIL>",
"SENDSOON_API_KEY": ""
}
}
}
}
Use only one top-level mcpServers object. Save the file, reopen Cursor, and confirm that sendsoon is enabled.
Codex
Open the user configuration file at ~/.codex/config.toml and add:
[mcp_servers.sendsoon]
command = "npx"
args = ["-y", "@sendsoon/ai"]
[mcp_servers.sendsoon.env]
SENDSOON_EMAIL_RECIPIENT = "<YOUR_EMAIL>"
SENDSOON_API_KEY = ""
Save the file and reopen Codex. Use /mcp to confirm that sendsoon is connected. In the desktop app, you can also check Settings > MCP servers.
Claude Code
Replace the placeholder and run:
claude mcp add --transport stdio --scope user --env SENDSOON_EMAIL_RECIPIENT=<YOUR_EMAIL> sendsoon -- npx -y @sendsoon/ai
Run /mcp in Claude Code and confirm that sendsoon is connected. To use an API Key, add --env SENDSOON_API_KEY=<SENDSOON_API_KEY> before the sendsoon server name.
Claude Desktop
Open the configuration file from Settings > Developer and add:
{
"mcpServers": {
"sendsoon": {
"command": "npx",
"args": ["-y", "@sendsoon/ai"],
"env": {
"SENDSOON_EMAIL_RECIPIENT": "<YOUR_EMAIL>",
"SENDSOON_API_KEY": ""
}
}
}
}
Save the file, fully quit and restart Claude Desktop, and confirm that sendsoon appears in the tool list.
Other MCP clients
For Windsurf, Cline, Continue, or another client that supports local stdio MCP servers, enter:
| Setting | Value |
|---|---|
| Transport | stdio |
| Command | npx |
| Arguments | -y @sendsoon/ai |
| Environment | The environment variables listed above |
Confirm that it works
Restart the client, open a new conversation, and enter:
Look up the location and ISP for 8.8.8.8.
The MCP connection works if the agent calls ip_lookup and returns the result.
To test email, enter:
Send a test email to <YOUR_EMAIL> with the subject “SendSoon MCP test” and the body “Configuration successful.”
The address must exactly match SENDSOON_EMAIL_RECIPIENT. Your client may ask for permission the first time a tool runs; approve it to continue.
Everyday examples
Look up information about 1.1.1.1.
Convert <YOUR_FILE_PATH> to Markdown and summarize the key points.
Send an email to <YOUR_EMAIL> with the subject “Meeting reminder” and the body “The meeting starts at 3 PM today.”
If the agent does not select a tool automatically, say:
Use the sendsoon MCP to complete this task.
You usually need this explicit instruction only on the first use.
Install Agent Skills
Agent Skills teach the agent when to call each MCP tool and how to handle error codes. Install the MCP server first, then add skills.
Claude Code (plugin marketplace)
/plugin marketplace add sendsoon/ai
/plugin install sendsoon-skills@sendsoon
Cursor / Claude (copy skills)
Copy any folder under skills/ into your agent skills directory:
| Client | Project path | User path |
|---|---|---|
| Cursor | .cursor/skills/<name>/ |
~/.cursor/skills/<name>/ |
| Claude Code | .claude/skills/<name>/ |
~/.claude/skills/<name>/ |
Available skills: email-basics, ip-lookup, markitdown.
License
See LICENSE.
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。