setup-mcp
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
ChineseInstall the Deepgram MCP Server
安装Deepgram MCP服务器
You are setting up Deepgram MCP integration for the user. Follow these steps:
你正在为用户设置Deepgram MCP集成,请遵循以下步骤:
Step 1: Check for the Deepgram CLI
步骤1:检查Deepgram CLI是否安装
Run (or , or on Windows) to check if the
Deepgram CLI is installed.
dg --versiondeepctl --versionwhere dg- If found: use the local CLI MCP () — this gives full access to Deepgram tools including transcription, text-to-speech, project management, and more.
dg mcp - If not found: use the hosted documentation MCP and offer to install the CLI.
运行(或,Windows系统下运行)来检查Deepgram CLI是否已安装。
dg --versiondeepctl --versionwhere dg- 如果已找到:使用本地CLI MCP()——这可让你完全访问Deepgram的各类工具,包括转录、文本转语音、项目管理等功能。
dg mcp - 如果未找到:使用托管文档MCP,并建议用户安装CLI。
Step 2: Detect the environment
步骤2:检测运行环境
Determine which AI coding tool the user is running. Check for:
- Claude Code — look for a directory in the project or user home
.claude/ - Cursor — look for a directory in the project root
.cursor/ - Windsurf — look for a directory in the project root
.windsurf/
If multiple are detected, or none are detected, ask the user which tool they want to configure.
确定用户正在使用的AI编码工具,检查以下内容:
- Claude Code——查看项目目录或用户主目录中是否存在文件夹
.claude/ - Cursor——查看项目根目录中是否存在文件夹
.cursor/ - Windsurf——查看项目根目录中是否存在文件夹
.windsurf/
如果检测到多个工具或未检测到任何工具,请询问用户想要配置哪一个。
Step 3: Ask about scope
步骤3:询问配置范围
Ask the user whether they want the MCP server configured:
- For this project only (recommended for team repos)
- Globally (available in all projects)
询问用户希望将MCP服务器配置为:
- 仅针对当前项目(推荐用于团队仓库)
- 全局配置(可在所有项目中使用)
Step 4: Install
步骤4:安装配置
If the Deepgram CLI (dg
) IS installed — use the local CLI MCP
dg如果已安装Deepgram CLI(dg
)——使用本地CLI MCP
dgClaude Code
Claude Code
sh
undefinedsh
undefinedProject scope
项目范围
claude mcp add deepgram --scope project dg mcp
claude mcp add deepgram --scope project dg mcp
User/global scope
用户/全局范围
claude mcp add deepgram dg mcp
undefinedclaude mcp add deepgram dg mcp
undefinedCursor
Cursor
Write or merge into the project's :
.cursor/mcp.jsonjson
{
"mcpServers": {
"deepgram": {
"command": "dg",
"args": ["mcp"]
}
}
}写入或合并到项目的文件中:
.cursor/mcp.jsonjson
{
"mcpServers": {
"deepgram": {
"command": "dg",
"args": ["mcp"]
}
}
}Windsurf
Windsurf
Write or merge into the project's :
.windsurf/mcp.jsonjson
{
"mcpServers": {
"deepgram": {
"command": "dg",
"args": ["mcp"]
}
}
}写入或合并到项目的文件中:
.windsurf/mcp.jsonjson
{
"mcpServers": {
"deepgram": {
"command": "dg",
"args": ["mcp"]
}
}
}Other tools
其他工具
Provide the MCP server command and let them configure manually:
- Transport: stdio
- Command:
dg - Args:
["mcp"]
提供MCP服务器命令,让用户手动配置:
- 传输方式:stdio
- 命令:
dg - 参数:
["mcp"]
If the Deepgram CLI is NOT installed — use the hosted docs MCP
如果未安装Deepgram CLI——使用托管文档MCP
Tell the user:
The Deepgram CLI isn't installed. You can either:
- Install the CLI (
) for full Deepgram tool access, then re-runpipx install deepctl/deepgram:setup-mcp- Use the hosted MCP for documentation queries right now
If they choose the hosted MCP (or want it alongside the CLI):
告知用户:
未检测到Deepgram CLI。你可以选择:
- 安装CLI(
)以获取完整的Deepgram工具权限,然后重新运行pipx install deepctl/deepgram:setup-mcp- 立即使用托管MCP进行文档查询
如果用户选择托管MCP(或希望同时使用CLI和托管MCP):
Claude Code
Claude Code
sh
undefinedsh
undefinedProject scope
项目范围
claude mcp add deepgram-docs --scope project --transport http https://api.dx.deepgram.com/kapa/mcp
claude mcp add deepgram-docs --scope project --transport http https://api.dx.deepgram.com/kapa/mcp
User/global scope
用户/全局范围
claude mcp add deepgram-docs --transport http https://api.dx.deepgram.com/kapa/mcp
undefinedclaude mcp add deepgram-docs --transport http https://api.dx.deepgram.com/kapa/mcp
undefinedCursor
Cursor
Write or merge into the project's :
.cursor/mcp.jsonjson
{
"mcpServers": {
"deepgram-docs": {
"type": "http",
"url": "https://api.dx.deepgram.com/kapa/mcp"
}
}
}写入或合并到项目的文件中:
.cursor/mcp.jsonjson
{
"mcpServers": {
"deepgram-docs": {
"type": "http",
"url": "https://api.dx.deepgram.com/kapa/mcp"
}
}
}Windsurf
Windsurf
Write or merge into the project's :
.windsurf/mcp.jsonjson
{
"mcpServers": {
"deepgram-docs": {
"type": "http",
"url": "https://api.dx.deepgram.com/kapa/mcp"
}
}
}写入或合并到项目的文件中:
.windsurf/mcp.jsonjson
{
"mcpServers": {
"deepgram-docs": {
"type": "http",
"url": "https://api.dx.deepgram.com/kapa/mcp"
}
}
}Other tools
其他工具
Provide the MCP server details and let them configure manually:
- Type: HTTP
- URL:
https://api.dx.deepgram.com/kapa/mcp
提供MCP服务器详情,让用户手动配置:
- 类型:HTTP
- URL:
https://api.dx.deepgram.com/kapa/mcp
Step 5: Confirm
步骤5:确认配置完成
After installation:
- Claude Code — run to activate immediately, no restart needed.
/reload-plugins - Cursor / Windsurf / Other — the user may need to restart or reload their tool.
Then tell the user:
The Deepgram MCP server is now configured.
If using the CLI MCP (), add:
dg mcpAvailable tools include transcription, text-to-speech, project management, and usage queries. Try asking to transcribe an audio file or convert text to speech.
If using the hosted docs MCP, add:
Your tool can now query Deepgram's full documentation directly — try asking about API parameters, voice agents, or model capabilities.
Link them to Deepgram Agentic Tools for more details.
安装配置完成后:
- Claude Code——运行即可立即激活,无需重启工具。
/reload-plugins - Cursor / Windsurf / 其他工具——用户可能需要重启或重新加载工具。
然后告知用户:
Deepgram MCP服务器已配置完成。
如果使用的是CLI MCP(),补充说明:
dg mcp你现在可以使用包括转录、文本转语音、项目管理和使用情况查询在内的各类工具。尝试请求转录音频文件或转换文本为语音。
如果使用的是托管文档MCP,补充说明:
你的工具现在可以直接查询Deepgram的完整文档——尝试询问API参数、语音Agent或模型能力相关问题。
将用户引导至Deepgram Agentic Tools以了解更多详情。
Troubleshooting
故障排查
If the MCP server fails to connect:
- For the CLI MCP: verify works and
dg --versionruns without errors. Update withdg mcporpipx upgrade deepctl.pip install --upgrade deepctl - For the hosted MCP: verify the URL is accessible.
https://api.dx.deepgram.com/kapa/mcp - To install the CLI:
pipx install deepctl
如果MCP服务器连接失败:
- 对于CLI MCP:验证可正常运行,且
dg --version启动无错误。使用dg mcp或pipx upgrade deepctl更新CLI。pip install --upgrade deepctl - 对于托管MCP:验证URL 可正常访问。
https://api.dx.deepgram.com/kapa/mcp - 安装CLI的命令:
pipx install deepctl