autowriter-mcp
If you are the rightful owner of autowriter-mcp and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.
Autowriter MCP Server is a token-saving Model Context Protocol server that facilitates AI-powered content generation by integrating Obsidian vaults with LMStudio, significantly reducing Claude token usage.
Autowriter MCP Server
A token-saving Model Context Protocol (MCP) server that coordinates automated writing workflows between Obsidian vaults and LMStudio. This server enables AI-powered content generation while dramatically reducing Claude token usage by generating content locally via LMStudio and saving directly to Obsidian vaults.
🚀 Key Features
- Token-Saving Architecture: Save 80-90% of Claude tokens by generating content locally via LMStudio
- Direct Obsidian Integration: Seamlessly integrates with Obsidian vaults and markdown files
- Automated Book Writing: Analyze book structure and generate missing sections automatically
- Batch Content Generation: Generate multiple sections efficiently in one operation
- Queue Management: Queue writing tasks and process them automatically
- Index Link Management: Automatically update Obsidian index files with generated content links
🛠 Prerequisites
- Python 3.11+: Required for running the MCP server
- LMStudio: Running locally on
http://localhost:1234
(or configurable URL) - Claude Desktop: With MCP support for connecting to the server
- Obsidian: For managing your writing vault (optional but recommended)
📦 Installation
Method 1: Using uvx (Recommended)
# Install and run directly with uvx
uvx autowriter-mcp '/path/to/your/obsidian/vault'
Method 2: Using pip
# Install from PyPI
pip install autowriter-mcp
# Run the server
autowriter-mcp '/path/to/your/obsidian/vault'
Method 3: Local Development
# Clone the repository
git clone https://github.com/infinitimeless/autowriter-mcp.git
cd autowriter-mcp
# Create virtual environment
python -m venv .venv
source .venv/bin/activate # On Windows: .venv\Scripts\activate
# Install dependencies
pip install -e .
# Run the server
autowriter-mcp '/path/to/your/obsidian/vault'
⚙️ Configuration
Command Line Options
autowriter-mcp [vault_path] [options]
Arguments:
vault_path Path to your Obsidian vault directory (required)
Options:
--index-file FILENAME Index file name (default: book_index.md)
--lmstudio-url URL LMStudio server URL (default: http://localhost:1234)
--version Show version information
--help Show help message
Claude Desktop Configuration
Add to your Claude Desktop MCP configuration:
For uvx installation:
{
"mcpServers": {
"autowriter-mcp": {
"command": "uvx",
"args": ["autowriter-mcp", "/path/to/your/obsidian/vault"]
}
}
}
For local development:
{
"mcpServers": {
"autowriter-mcp": {
"command": "/path/to/autowriter-mcp/.venv/bin/python",
"args": ["-m", "autowriter_mcp.server", "/path/to/your/obsidian/vault"]
}
}
}
🎯 Usage
1. Prepare Your Obsidian Vault
Create an index file (default: book_index.md
) in your vault with your book structure:
# My Book Title
## Chapter 1: Introduction
## Chapter 2: Getting Started
## Chapter 3: Advanced Topics
## Chapter 4: Conclusion
2. Available MCP Tools
analyze_book_structure
Analyze your vault structure and identify missing content sections.
generate_and_save_section
🚀 Token-Saving: Generate content locally via LMStudio and save directly to vault.
batch_generate_sections
🚀 Ultra Token-Saving: Generate multiple sections in one operation.
process_writing_queue
🚀 Auto Token-Saving: Process entire writing queue automatically.
request_content_generation
Add content generation requests to the queue for later processing.
update_index_links
Update your index file with Obsidian links to newly created content.
get_writing_status
Get current progress and queue status.
🏗 Architecture
Token-Saving Design
The server is specifically designed to minimize Claude token usage:
- Local Generation: Content is generated by LMStudio, not Claude
- Direct File Writing: Content is saved directly to vault files
- Metadata Only: Claude only receives generation metadata, not content
- Batch Processing: Multiple sections can be generated in one operation
📄 License
This project is licensed under the MIT License - see the file for details.
🙏 Acknowledgments
- Built on the Model Context Protocol
- Integrates with LMStudio for local AI generation
- Designed for Obsidian vault management
- Compatible with Claude Desktop MCP support
🚀 Save Claude tokens while accelerating your writing workflow with autowriter-mcp!