Trae Agent: FULLY FREE AI Coder + Opensource! BYE Gemini CLI & ClaudeCode? (Opensource)

Key Concepts

• Trey Agent: An open-source, MIT-licensed AI agent for software engineering tasks.
• CLI (Command Line Interface): A text-based interface for interacting with the AI agent.
• Large Language Models (LLMs): AI models like Claude Sonnet, Opus, and OpenAI models used by Trey Agent.
• Modular Architecture: Trey Agent's design, allowing for customization and experimentation.
• Tools: Specific functionalities within Trey Agent, such as editing files or executing shell commands.
• Open Router: A service providing a free, rate-limited API key for accessing LLMs.
• Trajectory: A saved record of the steps taken during an AI agent's execution.

Trey Agent: An Overview

Trey Agent is a new open-source tool developed by ByteDance, designed as an interactive CLI-based AI agent for automating software development tasks. It's built with Python and uses a modular architecture, making it customizable and suitable for research. Trey Agent is positioned as a free alternative to tools like Cursor and Windsurf. It recently topped the Sway Bench verified leaderboard with a score of 75.20%.

Installation and Setup

1. Prerequisites: Python (3.12+), Git, and an API provider (Open Router, OpenAI, or Anthropic).
2. Cloning the Repository: Use git clone [repository link] to download the Trey Agent code.
3. Navigating to the Directory: Use cd tray-agent to enter the project folder.
4. Installing Dependencies: Use the uv sync command to install the necessary packages.
5. Setting the API Key: Use the export command or modify the config file to set your API key. The config file also allows customization of parameters like max tokens and temperature.
6. Starting the Agent: Run the tray-cli command to launch the interactive mode.

Using Trey Agent

• Basic Task Execution: Use tray-cli run [prompt] to execute tasks.
• Specifying Providers and Models: Include parameters to select the desired provider (e.g., Open Router) and model.
• Working Directory: Set the path for the agent to operate within.
• Available Tools:
  • Edit Tool: Create, edit, view, and manipulate files.
  • Shell Command Execution: Run shell commands and scripts.
  • Sequential Thinking: Enables multi-step reasoning.
  • Task Completion Signaling: Indicates when a task is finished.
• Modes: Status (agent information), help (available commands), clear, exit.

Example Use Cases

• Creating a "Hello, World!" Script: The agent can generate a simple Python script and save the execution trajectory to a JSON file.
• Developing a Mind Super Game: The agent can create a functional game with different difficulty modes and features like animations.

Comparison with Other Tools

• Trey Agent vs. Cloud Code/Gemini CLI: Trey Agent can execute commands, edit files, and automate workflows using modular tools like bash, unlike Cloud Code, which primarily functions as a chat assistant for code generation. Trey Agent performs real actions in your system and supports multi-step reasoning.

Outskill Sponsorship

The video is sponsored by Outskill, which is hosting a free 3-day AI mastermind event. This 16-hour live sprint aims to help participants become confident in using AI for various tasks, including building agents, automating work, creating videos, and launching AI startups. The event is taught by experts from Microsoft, Amazon, and Samar AI. Participants also receive bonuses like a prompt bible, a mind map, and an AI toolkit.

Conclusion

Trey Agent is a promising open-source CLI agent for software engineering automation. Its modular design and ability to execute commands directly in the terminal make it a valuable tool for developers. While the UI could be improved, the core functionality and extensibility of Trey Agent are noteworthy. The presenter encourages viewers to explore the tool and provide feedback.