Dropbox Dash MCP Server

Dropbox Dash MCP Server

Exposes Dropbox Dash search and file metadata via STDIO, enabling users to search across company content and fetch detailed file metadata and content.

Category
访问服务器

README

Dropbox Dash MCP Server

An MCP server that exposes Dropbox Dash search and file metadata via STDIO using the Python MCP server library (fastmcp). Authenticate with Dropbox, then search across all company content and fetch detailed file metadata and content.

Tools Implemented

  • dash_get_auth_url

    • Summary: Start Dropbox OAuth; returns the authorization URL.
    • Args: none
    • Returns (text): A short instruction message followed by the URL.
    • Notes: Use when not yet authenticated or when a token has expired. After approval in the browser, call dash_authenticate with the one-time code. Tokens are stored securely in your system keyring, so this is typically a one-time setup.
  • dash_authenticate

    • Summary: Complete OAuth using the one-time authorization code.
    • Args:
      • auth_code (string, required)
    • Returns (text): Account display name and email on success; a human-readable error on failure.
    • Notes: Persists a token for subsequent tool calls. Typically only needed once until the token expires or is revoked.
  • dash_company_search

    • Summary: Search company content indexed by Dropbox Dash.
    • Args:
      • query (string, required) — search text
      • file_type (string or null, optional) — one of: document, image, video, audio, pdf, presentation, spreadsheet; or null for no filter. Default: null. The value document is also treated as no filter.
      • connector (string or null, optional) — filter by connector source. Common connectors include confluence, dropbox, github, gmail, gong, google_calendar, google_drive, jira, microsoft_365, microsoft_teams, slack, workday, zoom, among others. Default: null for no connector filter.
      • start_time (string or null, optional) — filter results modified after this datetime (ISO 8601 format, e.g., 2025-10-30T16:24:12.071Z). Default: null for no start time filter.
      • end_time (string or null, optional) — filter results modified before this datetime (ISO 8601 format, e.g., 2025-10-31T16:24:12.071Z). Default: null for no end time filter.
      • max_results (integer, optional) — default 20, range 1..100
    • Returns (text): A formatted list of results. Each result includes labeled fields such as UUID:, Type:, URL:, Preview:, Description:, File Type:, MIME Type:, Source:, Creator:, Last Modified By:, Updated:, Source Updated:, Relevance:, Source ID:. Results are separated by a divider line.
    • Errors: Human-readable messages for invalid parameters or missing authentication.
  • dash_get_file_details

    • Summary: Fetch detailed metadata (and optional content snippet) for a result UUID.
    • Args:
      • uuid (string, required) — UUID from search results
    • Returns (text): A summary with labeled fields (Title, Link, Updated, Source Updated, MIME Type, Source, Creator, Last Modified By). Media sections are included when present (Video/Image metadata). Content, when available, is shown with a MIME type and may be truncated to ~20,000 characters.
    • Errors: Human-readable messages for missing authentication or unknown UUID.

First-Time (or Re-Auth) Flow

If the user is not yet authenticated (or the token has expired):

  1. dash_get_auth_url → open the URL and approve access.
  2. dash_authenticate(auth_code) → store the token.
  3. Proceed with dash_company_search(...) and dash_get_file_details(uuid).

Prerequisites

Before installing and running the MCP server, you need to create a Dropbox app to obtain API credentials:

  1. Go to dropbox.com/developers/apps
  2. Click on Create app
  3. Select Scoped access as the API type
  4. Choose Full Dropbox access
  5. Give your app a name
  6. After creating the app, go to the Permissions tab and enable:
    • files.metadata.read
    • files.content.read
  7. Take note of the App key value from the Settings tab (you'll need this for configuration)

This credential will be used as APP_KEY in the installation steps below.

Requirements

  • Python 3.10 or higher
  • Dropbox Dash API credentials (App key)
  • Network access to Dropbox APIs

Installation

Clone the repository:

git clone https://github.com/dropbox/mcp-server-dash
cd mcp-server-dash

Install uv for virtual environment and dependency management:

macOS (Homebrew)

brew install uv

macOS/Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows (PowerShell):

irm https://astral.sh/uv/install.ps1 | iex

Set up the virtual environment and install dependencies

uv sync

Provide credentials (environment or .env)

export APP_KEY=your_dropbox_app_key
# or create a .env file with APP_KEY
# e.g., copy the example file:
cp .env.example .env

Usage

Running the MCP Server in STDIO mode

uv run src/mcp_server_dash.py

Authenticate via tools: call dash_get_auth_url, then dash_authenticate with the code. The access token is securely stored in your system's keyring (macOS Keychain, Windows Credential Manager, or Linux Secret Service).

Token Management

Tokens are stored securely in your system keyring under the service name mcp-server-dash or in ~/.mcp-server-dash/dropbox_token.json

To clear a stored token, use:

uv run src/mcp_server_dash.py --clear-token

Common MCP Server Configuration

For most MCP clients, including Claude and Cursor, you need to insert the below JSON configuration into a specific configuration file. See the specific instructions for Claude and Cursor below.

Important: Update the configuration below with the path to your installation and with your APP_KEY.

MCP Server Configuration:

{
  "mcpServers": {
    "Dropbox Dash Search": {
      "command": "uv",
      "args": [
          "--directory",
          "/path/to/mcp-server-dash/",
          "run",
          "src/mcp_server_dash.py"
      ],
      "env": {
        "APP_KEY": "your_dropbox_app_key"
      }
    }
  }
}

Security Note: For better security, consider using environment variables or a .env file with restrictive permissions (excluded from version control) instead of placing credentials directly in the MCP config file.

Using Claude as the Client

  • Open Claude Desktop → Settings → Developer → Local MCP Servers → Edit Config
  • Add the JSON MCP Server configuration shown above.
  • Restart Claude Desktop after saving the config.

Using Cursor as the Client

  • Press Cmd+Shift+P (macOS) or Ctrl+Shift+P (Windows) to open the Command Palette
  • Type "View: Open MCP Settings"
  • Add the JSON MCP Server Configuration to the mcp.json file as instructed.

Using Goose as the Client

  • Select Extensions → Add custom extension
  • Fill out the form:
    • Extension Name: Dropbox Dash Search
    • Type: STDIO
    • Description: Provide company context to your workflows
    • Command: uv --directory /path/to/mcp-server-dash/ run src/mcp_server_dash.py
    • Environment:
      • APP_KEY: Your Dropbox Client ID
  • Click Add Extension

Development

Install dev tools (ruff, black, mypy, pytest, coverage) using uv:

uv pip install -e ".[dev]"

Lint, Format, Type-check

Run checks (via uv):

# Lint (imports, style, bugbear, etc.)
uv run ruff check .

# Format (apply changes)
uv run black .

# Type-check
uv run mypy src

Tip: use uv run ruff format . (or uv run black .) to auto-format, and uv run ruff check --fix . to apply safe autofixes.

Testing & Coverage

Run the test suite (quiet mode with coverage summary) using uv:

uv run pytest
# or explicitly with coverage flags
uv run pytest -q --cov=src --cov-report=term-missing

Debugging

You can inspect and debug the server with the Model Context Protocol Inspector:

npx @modelcontextprotocol/inspector uv run src/mcp_server_dash.py

Ensure APP_KEY is set in your environment or .env before running the inspector.

License

Apache License 2.0

Copyright (c) 2025 Dropbox, Inc.

See LICENSE for details.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选