dropbox-transcripts-mcp

An MCP server that indexes plain-text podcast transcripts stored in Dropbox and makes them searchable inside Claude Code. Syncs automatically as new transcripts are added.

What it does

• Caches all .txt transcripts from a Dropbox folder into a local SQLite database with full-text search (FTS5)

• Polls Dropbox every N hours and indexes new or changed files automatically

• Exposes four tools to Claude Code: list, retrieve, search, and manual sync

Related MCP server: ClaudeHistoryMCP

Prerequisites

• Python 3.10+

• uv (brew install uv on macOS)

• A Dropbox account with your transcripts in a folder (default: /Podcasts/Lenny)

Setup

1. Create a Dropbox app

1. Go to https://www.dropbox.com/developers/apps

2. Click Create app

3. Choose Scoped access and Full Dropbox

4. Name it anything (e.g. transcripts-mcp)

5. Under the Permissions tab, enable:

   • files.metadata.read

   • files.content.read

6. Copy the App Key and App Secret from the Settings tab

2. Run the auth setup

uvx --from git+https://github.com/YOUR_USERNAME/dropbox-transcripts-mcp dropbox-transcripts-setup

This walks you through the OAuth flow and prints your DROPBOX_REFRESH_TOKEN. It will also print the exact MCP config block to paste into your Claude Code settings.

3. Add to Claude Code

Edit ~/.claude/settings.json and add:

{
  "mcpServers": {
    "transcripts": {
      "command": "uvx",
      "args": [
        "--from", "git+https://github.com/YOUR_USERNAME/dropbox-transcripts-mcp",
        "dropbox-transcripts-mcp"
      ],
      "env": {
        "DROPBOX_APP_KEY": "your_app_key",
        "DROPBOX_APP_SECRET": "your_app_secret",
        "DROPBOX_REFRESH_TOKEN": "your_refresh_token",
        "DROPBOX_FOLDER_PATH": "/Podcasts/Lenny"
      }
    }
  }
}

The first time Claude Code starts the server it will sync all transcripts from Dropbox. Subsequent syncs happen automatically in the background every 6 hours (configurable).

Configuration

All configuration is via environment variables:

File naming convention

Transcript files should be named Guest Name.txt. The filename (without .txt) becomes the episode identifier used by get_episode and shown in list_episodes.

Examples: Adam Fishman.txt, Elena Verna 2.0.txt

Available tools

License

MIT