Skip to content
Get packs

Trae Agent

An open-source command-line AI agent that carries out software engineering tasks using LLMs.

GitHub Download source
Stars
12.2k
Forks
1.4k
Licence
MIT
Language
Python
Last update
9 months ago
Type
CLI

What it is

Trae Agent is a command-line tool from ByteDance that takes plain-language instructions, such as fixing a bug or adding tests, and works through them using an LLM and a set of tools for editing files and running bash commands. It supports several model providers, including OpenAI, Anthropic, Google Gemini, OpenRouter, Ollama and Doubao. The project describes itself as a transparent, modular design aimed at researchers and developers who want to study or extend agent architectures.

  • Run a one-off task such as creating a script with trae-cli run "..."
  • Use interactive mode to iterate on a codebase conversationally
  • Run tasks inside a Docker container using --docker-image or --dockerfile-path
  • Record a trajectory file of agent actions for debugging or analysis

Great for

  • Developers who want a scriptable terminal agent that works on a local project directory
  • Researchers studying or modifying LLM agent architectures, as the README positions it
  • People who want to switch between model providers, including local models through Ollama
  • Users comfortable editing YAML config and managing their own API keys

Not ideal for

  • Users who want a polished, ready-made IDE experience, since the README describes only a CLI and an interactive terminal mode
  • Teams that need a stable release, because the README says the project is still being actively developed and no release is listed
  • People without an API key or local model, since the tool requires one to run
Step-by-step guide: how to use Trae AgentInstall, first run, examples, tips and FAQs
OpenHands vs Trae AgentSWE-agent vs Trae Agent

How to get it

  1. 1

    Clone and build from source

    You'll need: Python 3.12+ · UV (https://docs.astral.sh/uv/) · An API key for your chosen model provider, or a local Ollama setup · Docker (only for Docker mode)

    Clone the repository

    git clone https://github.com/bytedance/trae-agent.git

    Enter the project folder

    cd trae-agent

    Install dependencies with UV

    uv sync --all-extras

    Activate the virtual environment

    source .venv/bin/activate
  2. 2

    Run it

    • Copy trae_config.yaml.example to trae_config.yaml and add your API key
    • Run trae-cli show-config to check the configuration
    • Run a task with trae-cli run "Create a hello world Python script"
  3. 3

    Get the most out of it

    • Use --provider and --model flags to switch models per task
    • Enable Lakeview in the YAML config for short summaries of agent steps
    • Configure MCP servers, for example the Playwright MCP server shown in the README
    • Use --trajectory-file to save a session for later review
    • Use Docker options such as --docker-image to isolate task execution

Downloads

Source code (.zip)main branch · straight from GitHubOfficial website & docswww.trae.ai/

Strengths

  • Multiple LLM providers are supported, including a local option via Ollama
  • Configuration is explicit in YAML, and trajectory recording helps with debugging
  • Released under the MIT license with a technical report on arXiv

Trade-offs

  • The README states the project is still under active development, so behavior may change
  • Setup requires managing API keys and provider configuration yourself
  • No release tag is listed, so you install from source rather than a versioned package

How it compares

ProjectStarsTypeLevelUpdatedBest for
trae-agent (this)12.2kCLIintermediate9 months agoDevelopers who want a scriptable terminal agent that works on a local project directory
OpenHands92.5kAppintermediate4 days agoDevelopers who want coding agents running on a server while their laptop is off
SWE-agent20.7kAI model / agentintermediate11 days agoResearchers studying autonomous coding agents or running SWE-bench experiments
deer-flow84.9kAI model / agentadvanced4 days agoDevelopers who want to experiment with multi-agent workflows on their own infrastructure
MetaGPT71.1kFrameworkintermediate10 months agoDevelopers experimenting with multi-agent LLM workflows
DeepSeek-Reasonix36kAppbeginner4 days agoDevelopers who want an AI assistant to make multi-step changes in a local project

Stats synced from GitHub. The assessment above was drafted with AI from the project's README on 2026-10-29 and may miss recent changes — check the README before relying on a command. MIT licence: free to use, modify and distribute, including commercially, as long as you keep the copyright notice.