Trae Agent is an LLM-based CLI tool designed for general software engineering tasks, enabling users to execute complex workflows via natural language. It targets researchers and developers seeking a transparent, modular, and extensible framework for AI agent development and experimentation.

How It Works

Trae Agent employs a modular architecture that supports multiple LLM providers (OpenAI, Anthropic, Doubao, Azure, OpenRouter) and a rich tool ecosystem including file editing, bash execution, and sequential thinking. Its design prioritizes research flexibility, allowing deep customization of agent behaviors, tool implementations, and workflow patterns, facilitating ablation studies and novel capability development.

Quick Start & Requirements

• Install via uv sync after cloning the repository.
• Requires Python 3.12+.
• API keys for chosen LLM providers (OpenAI, Anthropic, OpenRouter, etc.) are necessary and can be set via environment variables or a configuration file.
• Official documentation: TRAJECTORY_RECORDING.md

Highlighted Details

• Multi-LLM support for OpenAI, Anthropic, Doubao, Azure, and OpenRouter.
• Features a "Lakeview" summarization for agent steps.
• Includes interactive mode for iterative development and trajectory recording for debugging.
• Offers a rich tool ecosystem with file editing, bash execution, and sequential thinking.

Maintenance & Community

The project is actively developed and welcomes community contributions. Guidelines for contributing and development are provided.

Licensing & Compatibility

Licensed under the MIT License, permitting commercial use and integration with closed-source projects.

Limitations & Caveats

The project is currently in an alpha stage, indicating potential instability and ongoing development. Priorities include expanding LLM provider support, enhancing the CLI, developing more tools, and implementing robust unit testing.