fdx-mcp-server
Enables reading and writing local files via MCP, supporting Bun, Deno, or Node.js runtimes.
README
Final Draft MCP Server
Description
fdx-mcp-server is a Model Context Protocol (MCP) server that lets AI agents read, analyze, and manipulate Final Draft screenplay files (.fdx). It exposes tools for parsing scene headings, character arcs, dual dialogues, SmartType dictionaries, pagination maps, script breakdowns, and more — effectively giving an LLM the ability to understand and edit screenplay structure.
Written in TypeScript, it runs on Bun or Deno using the stdio transport protocol, making it suitable for integration with any MCP-compatible client such as Claude Desktop.
Table of Contents
Installation
Prerequisites: Node.js
Node.js is required for Option B (Global NPM Install). If you plan to use that option, install it first using one of these methods:
- Windows / macOS: Download the LTS installer from nodejs.org and run it.
- Linux (apt):
curl -fsSL https://deb.nodesource.com/setup_lts.x | sudo -E bash - && sudo apt-get install -y nodejs - Homebrew (macOS / Linux):
brew install node
Verify the installation by running node --version and npm --version in your terminal.
Option A — Direct from GitHub (Bun or Deno)
If you have Bun, Deno, or both installed, you can run the server directly from raw.githubusercontent.com without installing it locally. No compilation or OS code-signing warnings needed.
⚠️ Rate-limit notice: This path fetches the server from
raw.githubusercontent.comevery time your MCP client starts. GitHub enforces an anonymous usage policy that limits unauthenticated requests to 60 per hour (across all ofgithub.comand its subdomains). If you exceed this limit, requests will be rejected with a403 Forbiddenerror until the window resets. Frequent restarts can trigger this — for heavy use, see Option B which caches everything locally.
Option B — Global NPM Install
If you prefer a traditional Node.js/NPM setup, install the package globally directly from GitHub. This downloads all dependencies locally so the server boots instantly and works 100% offline.
npm install -g github:dmongrel/fdx-mcp-server
To update later: npm update -g fdx-mcp-server
Usage
Add a configuration block to your MCP client's config file (e.g., claude_desktop_config.json or .mcp.json). Select the entries that apply to you:
Using Bun only:
{
"mcpServers": {
"fdx-mcp-server-bun": {
"command": "bun",
"args": ["run", "https://raw.githubusercontent.com/dmongrel/fdx-mcp-server/main/src/index.ts"]
}
}
}
Using Deno only:
{
"mcpServers": {
"fdx-mcp-server-deno": {
"command": "deno",
"args": ["run", "--allow-env", "--allow-read", "--allow-write", "https://raw.githubusercontent.com/dmongrel/fdx-mcp-server/main/src/index.ts"]
}
}
}
Using both Bun and Deno:
{
"mcpServers": {
"fdx-mcp-server-bun": {
"command": "bun",
"args": ["run", "https://raw.githubusercontent.com/dmongrel/fdx-mcp-server/main/src/index.ts"]
},
"fdx-mcp-server-deno": {
"command": "deno",
"args": ["run", "--allow-env", "--allow-read", "--allow-write", "https://raw.githubusercontent.com/dmongrel/fdx-mcp-server/main/src/index.ts"]
}
}
}
Using global NPM install:
{
"mcpServers": {
"fdx-mcp-server": {
"command": "fdx-mcp-server",
"args": []
}
}
}
Features
Key capabilities exposed by fdx-mcp-server:
- Document lifecycle — open, save, reload, create new
.fdxfiles; manage server-side document cache. - Scene analysis — parse scene headings (INT./EXT., location, time of day), extract scene index and properties, compute script stats and page maps.
- Character tracking — retrieve character lists, extension metadata (V.O., O.S.), per-character scene appearance counts, and arc beats across scenes.
- Dual dialogue support — read and create side-by-side dialogue blocks.
- SmartType dictionaries — manage characters, extensions, locations, transitions, scene intros, times of day, spell-check lists, and paragraph types.
- Formatting & styling — query and edit element settings (fonts, indentation, spacing) for every paragraph type; manage header/footer content.
- Title page management — read and write title, author, contact block, copyright, and based-on credits.
- Script breakdowns — generate full production breakdown reports (props, vehicles, camera, cast) as text or HTML/PDF.
- Macro system — query macro aliases and their activation scopes.
- Search & navigation — find paragraphs by text, list sections and section contents, retrieve revision colors and display board data.
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。