When you start a project in ThinkRail, you have two options: work directly in the project folder or create a new workspace. Working directly on the project folder means that all changes will immediately affect the current git branch. This is straightforward but can also be unsafe. Unsaved changes may be removed by the AI agent. Saved changes could still be overwritten if the agent has permissions to perform certain git operations. Moreover, while you still can use more than one AI agent at once when working in the project folder, you risk that they will get in the way of each other. A workspace might be a solution to both those issues.
Create an isolated workspace
To create a workspace, click the “Start building” button. In the new window, select the git branch for your workspace and provide a description. This description serves as both the human-friendly workspace name and the initial prompt for the AI agent. (but note that the AI agent may later rename the workspace for conciseness). Once created, the workspace launches a new chat with an AI agent, which immediately begins working on the task.
A key feature of workspaces is their isolation. After the AI agent makes changes, opening the project in your IDE
will still show the original version. This is because the updated code resides in a separate worktree directory under
the ~/.thinkrail, for example ~/.thinkrail/worktrees/<project-name>/workspace-1. All changes made by the AI agent
are safely contained and will only be propagated to the parent branch when you request it.
ThinkRail uses git worktree functionality to create and manage these isolated environments. You can read more about
git worktrees here.
Work safely on many tasks at the same time
On top of that, while one AI agent works on a complex task, you can create another workspace from the same branch and assign a different task to a second AI agent. For instance, one agent can implement a change that does not modify the app’s behavior while another updates project specifications; these two tasks do not interfere with each other. Since workspaces are isolated, there is no need to wait for one task to complete before starting the other. Instead, you can run them concurrently and save time.
Once all tasks are complete, you can merge the changes back to the parent branch. If your workspaces originated from the main branch, you can now tell the AI agent to create a pull request (there is a skill for that). If they branched from a feature branch instead, you can instruct the AI agent to push the changes to that branch or simply do it yourself from the terminal.
That is all for now. To try out ThinkRail, visit our main webpage and follow the installation instructions. You can also browse our GitHub repository, take our survey, and join our Discord server to say “hi!” and let us know what you think. Your feedback is greatly appreciated.
Happy developing!
ThinkRail is backed by JetBrains, leveraging their expertise in building developer tools that are both powerful and intuitive.