close
Skip to content

realcarsonterry/Terry-Carson-swe-agent

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 

Repository files navigation

Terry-Carson-swe-agent

Terry chen's SWE Agent powered by GitHub Copilot Pro

A Minimal, Ready-to-Run SWE Agent — Designed specifically for GitHub Copilot Pro users (especially for students who get education ceritification LOL).

Powered by your Copilot Pro subscription, this agent acts like a real software engineer: it automatically clones repositories, explores code, fixes GitHub Issues, and generates a ready-to-apply patch.

Python License Powered by


✨ Features

  • Zero-config Copilot Pro support via LiteLLM (uses gpt-4o, Claude, etc. with your existing Copilot Pro quota)
  • Fully autonomous Issue fixing: runs ls, cat, grep, sed -i, tests, and more until the bug is fixed
  • Clean output: automatically creates final_patch.diff that you can apply with git apply
  • Safe by design: every task runs in a temporary folder that is deleted when finished
  • Perfect Windows support (PowerShell + CMD optimized)
  • Extremely simple code: only ~150 lines, easy to read, modify, and learn from

🚀 Quick Start

1. Clone this repository

git clone https://github.com/YOUR-USERNAME/my-swe-agent.git
cd my-swe-agent

2. Create and activate a virtual environment (recommended)

# Windows PowerShell
.venv\Scripts\Activate.ps1

# Windows CMD
.venv\Scripts\activate.bat

# macOS / Linux
source .venv/bin/activate

###3. Install dependencies

pip install -r requirements.txt

###4. Run the agent (first run will open browser for authorization)

python swe_agent.py https://github.com/pallets/flask.git "Fix a simple bug: when input is empty, the homepage should show a welcome message instead of throwing an error"

🛠️ How It Works

1. Clones the target repository into a temporary folder

2. Feeds the Issue description to the Copilot Pro model

3. Runs a ReAct-style loop:

•  Thought: thinks about the next step
•  Command: outputs exactly one bash command

4. Executes the command and feeds the output back to the model

5. Repeats until the model decides the fix is complete, then generates the final patch

The entire process is fully autonomous — just like having a junior software engineer working for you.

🤝 Welcome Friends from All Over the World to Contribute and Use!

No matter which country you are from or what language you speak, you are warmly welcome!

• Star this repository — it gives us the biggest motivation

• Fork and submit Pull Requests (improve prompts, add Docker support, local Ollama, multi-agent mode, auto-create PRs, etc.)

• Open Issues if you encounter any problems (Windows, authentication, specific models, etc.)

• Use Discussions to share real cases where you successfully fixed Issues with this agent

We look forward to your arrival .Let’s build the future more like the future together!

About

Terry chen's SWE Agent powered by GitHub Copilot Pro

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages