scrivener

Scrivener MCP Server provides a comprehensive interface for Claude AI to interact with Scrivener projects. It exposes a suite of MCP tools that allow you to open Scrivener projects, read and write documents (including RTF-formatted content), manage metadata, and navigate the Scrivener binder structure. The server also includes advanced analysis capabilities such as writing quality indicators, readability metrics, and various content enhancement tools, all designed to work in conjunction with Claude to improve drafting, editing, and organization workflows within Scrivener projects. With embedded memory concepts and project-scoped memory features, it enables persistent context across sessions without relying on external services like Redis.

Prerequisites:

• Node.js installed on your system (recommended LTS version)
• npm or yarn available

Installation steps:

1. Install the MCP server globally via npm:

npm install -g scrivener-mcp

2. Start the MCP server (the CLI exposes the MCP endpoint):

scrivener-mcp

3. Optional manual setup for Claude Desktop integration:

• If automatic configuration fails, run setup to generate the CLI configuration:

# Run the setup script
npx scrivener-mcp setup

• Or manually add to claude_desktop_config.json as shown:

{
  "mcpServers": {
    "scrivener": {
      "command": "npx",
      "args": ["scrivener-mcp"]
    }
  }
}

4. Uninstall if needed:

npm uninstall -g scrivener-mcp

Notes:

• The server ships with an embedded queue so Redis is not required for normal operation.
• Automatic Claude Desktop configuration is supported; restart Claude after installation to apply.
• The default MCP config exposes a server named 'scrivener' that runs via npx scrivener-mcp. You can adjust the mcpServers entry if you host differently.
• Ensure you have appropriate Scrivener access rights and the Scrivener application installed where the MCP server can interface with .scriv project files.
• If you utilize advanced features (memory persistence, HHM semantic search, or AI-enhanced editing), verify your environment supports any optional dependencies required for those modules.