Auto-execution CLI tool for LLM-powered task automation using local Ollama processing.

This project enables automatic tool execution using local LLM processing. The LLM automatically:

• Analyzes user requests and determines required tools
• Executes tools automatically without manual intervention
• Continues conversation with tool results
• Handles simple tasks through focused tool usage

• Auto-Execution: LLM automatically selects and executes tools
• Local AI Processing: Uses Ollama for 100% local AI processing
• Tool-Based Architecture: 8 specialized tools for automation
• Interactive CLI: Inquirer-based command line interface
• Context Management: Persistent conversation history and session management
• Intelligent Tool Selection: LLM chooses appropriate tools automatically

# Clone the repository
git clone https://github.com/NeaByteLab/Ollama-API-Tools
cd Ollama-API-Tools

# Install dependencies
npm install

# Install Ollama (if not already installed)
# Visit: https://ollama.ai or https://ollama.com

• Node.js 18+
• Ollama running locally
• Granite 3.2:8b model (or compatible model)

npm start
# or
node CLI.js

node CLI.js info

1. User Input: You describe what you want
2. LLM Analysis: LLM analyzes request and selects tools
3. Auto-Execution: Tools execute automatically
4. Result Processing: Results are added to conversation
5. Continued Conversation: LLM responds with results

Edit CLI.js line 40 to change the default model:

const defaultModel = 'granite3.2:8b' // Change this

• granite3.2:8b (default)
• qwen2.5-coder:7b
• llama3.2:8b
• Any Ollama-compatible model

# List directory contents
You: list files in current directory

# Read file contents  
You: read package.json

# Search for patterns
You: find all files containing "import"

# Edit files
You: add a comment to the main function

# Check running services
You: check if redis is running

# Execute commands
You: run npm install

# System information
You: show system info

# Fetch web content
You: get content from https://example.com

# Basic file operations
You: read package.json and show dependencies

# Simple search
You: find all files containing "import" in src folder

• Intelligent Tool Selection: LLM automatically chooses appropriate tools
• Single Tool Execution: Simple, focused tool usage per request
• Error Handling: Automatic error reporting and fallback
• Context Awareness: Maintains conversation context across tool executions
• Result Integration: Tool results automatically feed back into conversation

Ollama-API-Tools/
├── CLI.js              # Main CLI application with auto-execution
├── package.json        # Dependencies and scripts
├── prompt/            # AI prompt configurations
│   ├── prompt.txt     # Main system prompt
│   └── message.json   # Message templates
├── tools/             # Tool implementations
│   ├── Registry.js    # Tool registry
│   ├── BashTool.js    # Command execution
│   ├── ReadTool.js    # File reading
│   └── ...           # Other tools
├── services/          # Service layer
│   └── Provider.js    # Ollama API integration
├── utils/             # Utilities
│   └── Context.js     # Session management
└── tests/             # Test files
    └── TestTools.js   # Tool testing

1. Create tool file in tools/ directory
2. Implement tool class with execute() method
3. Add to Registry.js exports
4. Update prompt.txt tool list

npm test
# or
node tests/TestTools.js

Ollama not running

# Start Ollama
ollama serve

Model not found

# Pull model
ollama pull granite3.2:8b

Tool execution errors

• Check file permissions
• Verify tool dependencies
• Review error messages in console

Auto-execution not working

• Check LLM model compatibility
• Verify tool schema definitions
• Review prompt configuration

MIT License - NeaByteLab