Trae Agent is an open-source autonomous coding agent framework developed by ByteDance. It provides an agentic execution harness designed to solve repository-level software engineering tasks, orchestrating large language models to inspect codebases, execute terminal commands, modify files, and iterate on solutions. Operating primarily via command-line workflows and server interfaces, Trae Agent runs an autonomous decision-action loop. It interacts with LLMs (such as OpenAI and OpenAI-compatible API endpoints), tracks tool invocation outputs across multi-step message histories, and executes workspace actions based on YAML-driven configuration profiles. What sets Trae Agent apart is its tight coupling with standard coding benchmarks. The repository includes integrated pipelines for running and evaluating tasks on SWE-bench, SWE-bench Live, and Multi-SWE-bench datasets, positioning it both as a task-solving CLI agent and as a framework for benchmarking agentic coding performance. Trae Agent is open-source and operates on a Bring-Your-Own-Key (BYOK) model, requiring developer-provided API credentials for the target LLM providers.
SWE-bench Evaluation Suite — Built-in harness to run and evaluate tasks against SWE-bench, SWE-bench Live, and Multi-SWE-bench datasets. - OpenAI-Compatible Backend — Connects to OpenAI API endpoints and compatible model providers for agent reasoning and tool calling. - Tool Execution & State Persistence — Automatically executes file and shell actions while preserving tool outputs inside the conversation context history. - YAML Configuration — Centralized YAML-based configuration for setting up agent parameters, model choices, and environment variables. - Modular Agent Core — Structured Python codebase separating the agent core, server modules, and evaluation harnesses.
Use cases
Use Case: Benchmark & Model Evaluation AI researchers and developers can run Trae Agent against SWE-bench and SWE-bench Live datasets to quantitatively measure how well specific LLMs solve real-world GitHub issues end-to-end.
Use Case: Automated Issue Resolution Developers can pass a repository problem description or bug report to Trae Agent to automatically locate relevant files, implement fixes, and verify code modifications.
Use Case: Agent Architecture Research Engineers building custom coding harnesses can use the Trae Agent framework as a baseline to test new agentic prompt strategies, context management mechanisms, and tool calling interfaces.
Who it is for
Trae Agent is designed for software engineers, AI researchers, and developer tooling teams who need a configurable CLI harness for autonomous coding tasks and benchmark evaluations. It is not intended for non-technical users looking for a simple click-and-run GUI editor without API configuration.
Install guide
Installation Clone the repository and install the required dependencies (standard Python setup): bash git clone https://github.com/bytedance/trae-agent.git cd trae-agent pip install -e .
First Run Configure your model credentials and YAML configuration: bash export OPENAI_API_KEY="your-openai-api-key" # Copy and adjust the default configuration template cp config.example.yaml config.yaml
Example Session Run an agent task or execute a benchmark evaluation run: bash python evaluation/run_evaluation.py --config config.yaml
Plandex — Plandex is an open-source terminal-based AI coding agent designed to handle complex, multi-file engineering tasks in large codebases. Created by…
OpenCode — OpenCode is an open-source AI coding agent developed by Anomaly that operates across terminal environments, desktop applications, and IDE extensions.…
OpenHands — OpenHands (formerly OpenDevin) is an open-source autonomous AI software development agent and orchestration platform. It provides an extensible runtime environment…
Subtask — Subtask is an open-source CLI agent harness and orchestration tool developed by zippoxer in Go. It is built to coordinate…
EchoCoding — EchoCoding is an open-source agent harness and audio-augmented developer tooling layer created by launsion-boop. Designed to run alongside AI coding…
pi-reflect — pi-reflect is an open-source agent harness and self-improvement extension for pi coding agents, developed by jo-inc. It inspects agent session…