AI Novel Prompter is a comprehensive writing tool designed to help novelists create consistent, well-structured prompts for AI writing assistants while managing story elements and improving prose through AI-powered assistance.
- Story Element Management: Characters, locations, codex entries, rules, and sample chapters
- Intelligent Prompt Generation: Optimized for ChatGPT and Claude with real-time preview
- Prose Improvement Engine: Iterative text refinement with customizable AI prompts
- Local Data Storage: All data saved locally with versioned storage system
- Claude Desktop Integration: Direct access to your story data through Claude Desktop
- Real-time Character/Location Access: Query and manage entities directly in Claude conversations
- Configurable Logging: Debug character loading issues with detailed logging
- Robust Data Validation: Prevents creation of incomplete or empty entities
# Clone the repository
git clone https://github.com/danielsobrado/ainovelprompter.git
cd ainovelprompter
# Install frontend dependencies
cd frontend
npm install
cd ..
# Development mode
wails dev
# Build for production
wails build- Build the MCP Server:
cd mcp
go build -o ainovelprompter-mcp.exe ..\cmd\mcp-server\main.go- Configure Claude Desktop (edit
claude_desktop_config.json):
{
"mcpServers": {
"ai-novel-prompter": {
"command": "C:\\path\\to\\ainovelprompter-mcp.exe",
"args": [
"--data-dir", "C:\\path\\to\\your\\story\\data",
"--log-level", "INFO"
],
"env": {}
}
}
}- Restart Claude Desktop to load the MCP server
| Argument | Short | Description | Default |
|---|---|---|---|
--data-dir |
-d |
Data directory path | ~/.ai-novel-prompter |
--log-level |
-l |
Logging verbosity | INFO |
--help |
-h |
Show help message | - |
- DEBUG: Detailed operation logging (for troubleshooting)
- INFO: Standard operational logging (recommended)
- WARN: Warnings and unexpected conditions only
- ERROR: Errors only
Development/Troubleshooting:
"args": ["--data-dir", "./test-data", "--log-level", "DEBUG"]Production:
"args": ["--data-dir", "/path/to/story/data", "--log-level", "INFO"]- Required Fields: All entities must have non-empty names and descriptions
- Length Constraints: Names (1-200 chars), Descriptions (1-2000 chars)
- Automatic Sanitization: Whitespace trimming and cleanup
- Error Prevention: Cannot create incomplete entities
- ✅ Character names and descriptions required
- ✅ Location names and descriptions required
- ✅ Codex entry titles and content required
- ✅ Automatic whitespace cleanup
- ❌ Empty or whitespace-only entities rejected
-
Prompt Generation Tab:
- Define task types (e.g., "Write Next Chapter", "Revise Chapter")
- Manage story elements (characters, locations, rules, codex)
- Generate AI-optimized prompts for ChatGPT or Claude
- Real-time token counting and preview
-
Prose Improvement Tab:
- Configure LLM providers (Manual, LM Studio, OpenRouter)
- Apply customizable improvement prompts
- Review and accept/reject AI suggestions
- Live diff preview of changes
- Configure LLM providers (Manual, LM Studio, OpenRouter)
Once configured, you can directly interact with your story data in Claude conversations:
# Get all characters
Show me my characters
# Search for specific characters
Find characters related to "magic"
# Get locations
What locations do I have in my story?
# Search locations
Show me locations in the forest
- React with TypeScript
- Tailwind CSS for styling
- shadcn/ui components
- Wails framework for desktop
- Go with Wails bindings
- MCP (Model Context Protocol) server
- Versioned JSON storage system
- Comprehensive validation layer
- JSON-RPC 2.0 protocol
- Stdio communication with Claude Desktop
- Configurable logging system
- Real-time data access
~/.ai-novel-prompter/
├── characters/
│ └── [character-id]/
│ ├── 2025-07-22T10-30-15.000+04-00.json
│ └── 2025-07-22T11-45-20.000+04-00.json
├── locations/
├── codex/
├── rules/
├── sample-chapters/
├── task-types/
└── prose-prompts/
- Version History: Complete audit trail for all changes
- Data Integrity: Individual files reduce corruption risk
- Scalability: Better performance with large datasets
- MCP Compatibility: Direct access through Claude Desktop
- Enable Debug Logging:
"args": ["--log-level", "DEBUG"]-
Check Claude Desktop Logs for detailed error information
-
Common Issues:
- Invalid timestamps in filenames
- Corrupted JSON files
- Permission problems
- Missing data directories
- Verify Server Binary: Ensure
ainovelprompter-mcp.exeis built with latest code - Check Configuration: Verify paths and arguments in
claude_desktop_config.json - Restart Claude Desktop: Required after configuration changes
- Review Logs: Look for startup errors and connection issues
- ✅ MCP Server Integration: Direct Claude Desktop access
- ✅ Enhanced Logging: Configurable debug levels with performance tracking
- ✅ Data Validation: Prevents incomplete entity creation
- ✅ Improved Character Loading: Robust timestamp parsing and error handling
- ✅ Command Line Interface: Flexible configuration options
- ✅ Frontend Validation: Immediate feedback with specific error messages
- ✅ Backend Validation: Server-side validation with detailed errors
- ✅ Automatic Sanitization: Whitespace trimming and cleanup
- ✅ Length Constraints: Proper limits for names and descriptions
Contributions are welcome! Please feel free to:
- Open issues for bugs or feature requests
- Submit pull requests for improvements
- Share feedback and suggestions
This project is licensed under the Attribution-NonCommercial-NoDerivatives (BY-NC-ND) license.
See: https://creativecommons.org/licenses/by-nc-nd/4.0/deed.en
- Repository: https://github.com/danielsobrado/ainovelprompter
- Wails Framework: https://wails.io/
- shadcn/ui Components: https://ui.shadcn.com/
- Model Context Protocol: https://modelcontextprotocol.io/
Last Updated: July 2025 - MCP Integration & Validation Release