Secret MCP
Enables AI assistants to write .env files with secrets stored locally, allowing search by name/description while keeping actual secret values private and never exposed to the AI.
README
Secret MCP
A desktop app for managing secrets with an MCP server that lets AI coding assistants write .env files without ever seeing the secret values.
Why?
Generating .env file is a pain, especially because there is no vibing out of it. But its an even stronger pain to leak secrets to AI coding assistants run on the cloud.
Features
- Desktop App: Simple window for managing secrets (name, description, value)
- MCP Server: Two tools for AI assistants:
search_secrets: Find secrets by name/description (never exposes values)write_env: Write secrets to.envfiles (values go straight to file, never to AI)
- Local Storage: All secrets stored locally in SQLite
- npm Package: Just
npx secret-mcp- no build required
Installation
Desktop App
Download from Releases or build from source:
npm install
npm run tauri build
MCP Server Setup
Add to your MCP client config:
"secret-mcp": {
"command": "npx",
"args": ["secret-mcp"]
}
Usage
- Open Secret MCP app
- Add your secrets (API keys, tokens, etc.)
- When coding with AI, it will automatically use
search_secretsandwrite_envto set up your.envfiles (Note: the name of the secret is the variable name in the.envfile)
MCP Tools
search_secrets
Search for secrets by name or description. Returns names and descriptions only - values are never exposed.
// Input
{ query: "openai" }
// Output
[
{ name: "OPENAI_API_KEY", description: "OpenAI API key" }
]
write_env
Write secrets to a .env file. Values go directly from your local database to the file - never passed through the AI.
// Input
{
keys: ["OPENAI_API_KEY", "DATABASE_URL"],
path: "/Users/you/project/.env"
}
// Output
"Successfully wrote 2 secret(s) to /Users/you/project/.env"
Data Storage
Secrets are stored locally:
- macOS:
~/Library/Application Support/secret-mcp/secrets.db - Linux:
~/.local/share/secret-mcp/secrets.db - Windows:
%APPDATA%/secret-mcp/secrets.db
Security
- Secret values never leave your machine (except to
.envfiles you specify) - MCP server only returns secret names and descriptions to the AI
.envfiles written with600permissions (owner read/write only)
Tech Stack
- Desktop: Tauri 2.0 + Svelte 5 + TypeScript
- MCP Server: Node.js + @modelcontextprotocol/sdk + better-sqlite3
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。