AI Agent Test

A lightweight, extensible agentic workflow system built with TypeScript and the AI SDK. This project serves as a testbed for implementing mini agentic flows with support for small-sized local LLMs (30B-80B parameters) and few tool integrations.

🌟 Features

• Agentic Architecture: Multi-iteration agent loop with tool calling capabilities
• Local LLM Support: Connect to any OpenAI-compatible API endpoint or Google Generative AI
• Extensible Tools: Built-in tools for file operations, bash execution, web search, and more
  • See src/tools/ for all available tools
• Interactive CLI: Real-time chat interface with streaming responses
• Debug Mode: Inspect conversation history and token usage
• Logging: Session logs are stored in ~/.ai/logs/ to track agent behavior and progress

🛠️ Prerequisites

• A local LLM server (e.g., LM Studio, Ollama) or cloud API key

📦 Installation

As a CLI tool (global)

npm install -g ai-agent-test
# Create ~/.ai/models.json with your configuration (see Configuration section below)

Then run:

ai

Development installation

npm install
# Create ~/.ai/models.json with your configuration (see Configuration section below)

⚙️ Configuration

Model Configuration

Create a JSON configuration file at ~/.ai/models.json with an array of model configurations:

[
  {
    "modelApiType": "openai",
    "modelName": "qwen3-coder-next",
    "apiBaseUrl": "http://localhost:1234/v1",
    "apiKey": "dummy",
  },
  {
    "modelApiType": "google",
    "modelName": "gemini-2.5-flash",
    "apiBaseUrl": "https://generativelanguage.googleapis.com/v1beta",
    "apiKey": "your_google_api_key_here",
  }
]

Start CLI with selected model (default is the first model in the array):

# option 1
ai 1 # to select the second model (Google Gemini)

# option 2
export AI_MODEL_INDEX=1
ai

# option 3
AI_MODEL_INDEX=1 ai

Additional Environment Variables

• BRAVE_API_KEY: (Required) API key for Brave Search for the internet search tool

CLI Commands

When installed globally, run ai to start the interactive CLI:

• Type your prompt and press Enter to start the agent loop
• exit or quit to terminate the session
• debug <n> to view last n messages (or all if no number provided)
• ESC key to interrupt the current agent iteration