An LLM-driven tool for analyzing Excel files and importing their data into SQLite databases. The AI agent handles all analysis, decision-making, and data processing while TypeScript provides simple execution tools.

• LLM-Driven Architecture: AI agent makes all decisions about parsing, data types, and import strategy
• Smart Column Naming: AI proposes meaningful SQL column names that preserve original meaning
• Excel Analysis: Examine file structure, sheets, headers, and data patterns
• Automatic Database Creation: Generate SQLite tables with appropriate data types
• Data Import: Import Excel data with intelligent column mapping
• Interactive Chat Interface: Natural language interaction with the AI agent

# Interactive mode - start a chat session tsx src/analyzeFile.ts <excel-file-path> <workspace-directory> # With initial prompt for direct testing tsx src/analyzeFile.ts <excel-file-path> <workspace-directory> "your prompt here"

# Interactive analysis tsx src/analyzeFile.ts "data/social-media-data.xlsx" "output" # Direct workflow with prompt tsx src/analyzeFile.ts "data/social-media-data.xlsx" "output" "Analyze this file and import all data into SQLite"

1. AI Agent Initialization: The LLM agent starts with access to simple TypeScript tools
2. File Analysis: The AI examines Excel files using tools to:
   • Get sheet information and structure
   • View raw data to understand layouts
   • Analyze headers and propose meaningful SQL column names
   • Determine appropriate data types and import strategies
3. Database Creation: The AI creates SQLite tables with intelligent column mappings
4. Data Import: The AI imports Excel data using the proposed column mappings

The tool creates the following files in the workspace directory:

• parserConfig.json - Parser configuration for the Excel file
• dataLoaderMetadata.json - Table configuration with column mappings
• data.db - SQLite database with the imported data

The AI assistant has access to these simple tools:

• getSheetsTool - Get list of sheets in the Excel file
• getRawDataTool - Get raw data from a specific sheet
• getParsedHeadersTool - Get parsed headers and propose meaningful SQL column names
• writeConfigTool - Save parser and table configuration
• createTableTool - Create SQLite table with column mappings
• importDataTool - Import Excel data into SQLite table

• Node.js with TypeScript support
• SQLite3
• Excel files (.xlsx format)

1. Open the project in VS Code
2. When prompted, click "Reopen in Container" or use Command Palette: "Dev Containers: Reopen in Container"
3. The container will automatically set up mise and install all dependencies

# Install mise (if not already installed) curl https://mise.run | sh # Install tools specified in mise.toml mise install # Install project dependencies pnpm install

• xlsx - Excel file reading
• sqlite3 - SQLite database operations
• umwelten - AI framework for agent interactions
• zod - Schema validation for tool inputs

This tool follows an LLM-driven architecture where:

• The AI agent makes all complex decisions (parsing, data types, column naming)
• TypeScript provides simple, focused tools for execution
• The LLM analyzes Excel structure and proposes meaningful SQL column names
• All analysis logic is handled by the AI, not hardcoded TypeScript

The project includes a Dev Container configuration that provides:

• Consistent Environment: Same Node.js and tool versions across all machines
• mise Integration: Automatic tool installation and management via mise
• VS Code Integration: Pre-configured extensions and settings
• GitHub Copilot Ready: Optimized for AI-assisted development

The dev container ensures that GitHub Copilot and other AI tools have access to the correct mise-managed environment.