Trae Agent is an LLM-based CLI tool for general software engineering tasks, supporting multiple providers, with a configurable YAML setup and trajectory logging.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Trae Agent is an LLM-based agent for general purpose software engineering tasks. It provides a CLI interface that can understand natural language instructions and execute software engineering workflows using various tools and LLM providers.
How it works
The project supports multiple model providers (OpenAI, Anthropic, Doubao, Azure, OpenRouter, Ollama, Google Gemini) and offers a rich tool ecosystem (file editing, bash execution, sequential thinking, etc.). It records agent actions in trajectories for debugging and analysis. Configuration is YAML-based with environment variable support.
Getting started
Installation is via cloning the repository and setting up a virtual environment:
git clone https://github.com/bytedance/trae-agent.git
cd trae-agent
uv sync --all-extras
source .venv/bin/activate
Configuration is done through trae_config.yaml.example copied to trae_config.yaml and edited with API keys and preferences:
agents:
trae_agent:
enable_lakeview: true
model: trae_agent_model
max_steps: 200
tools:
- bash
- str_replace_based_edit_tool
- sequentialthinking
- task_done
model_providers:
anthropic:
api_key: your_anthropic_api_key
provider: anthropic
openai:
api_key: your_openai_api_key
provider: openai
Usage commands include basic task execution and interactive mode:
trae-cli run "Create a hello world Python script"
trae-cli show-config
trae-cli interactive
Recent releases
- none
Traction
Stars: 11975 Forks: 1327 Open issues: 158
License
MIT
Caveats
License: MIT, as indicated in README; no age note beyond created date 2025-06-13 and last_push 2026-02-05.