JS Agent is a framework for building AI agents using JavaScript and TypeScript, designed to simplify the creation of reliable and robust AI applications. It offers composable building blocks and tooling for developers, enabling faster prototyping and development of complex agent behaviors.

How It Works

The framework emphasizes composability, extensibility, and production readiness. It provides a typed, functional programming approach for combining components like LLM providers, tools, and agent loops. Key features include configurable agent run properties, observation capabilities for logging and UI integration, cost tracking for LLM calls, and the ability to integrate multiple LLM models within a single agent. Agents can be managed via an HTTP API, supporting features like tool execution within Docker containers and various agent loop strategies such as BabyAGI-style task planning.

Quick Start & Requirements

• Install via npm: npm install js-agent
• Requires Node.js.
• OpenAI API key and potentially Google Search API keys (for specific examples) are needed for LLM interactions.
• See js-agent.ai for documentation and examples.

Highlighted Details

• Supports OpenAI models (GPT-4, GPT-3.5-turbo, text-davinci-003, text-embedding-ada-002).
• Includes built-in tools for file I/O, CLI commands, web scraping, and user interaction.
• Offers flexible JSON action formatting and prompt templating for chat and text models.
• Features agent HTTP server for remote management and observation.

Maintenance & Community

The project is noted as not actively developed, with a recommendation to check out ModelFusion as a successor. The primary contributor is lgrammel.

Licensing & Compatibility

The README does not explicitly state a license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The project is explicitly stated as not actively developed and is considered an early library for testing ideas. Users are directed to ModelFusion for more current development.