Skip to main content
Setting up the task runner is optional, but it will certainly improve results. It helps cto.new understand how to run and check your code.This helps cto.new write code that passes any pre-commit checks and pipeline steps, and ensures the code does what you expect.
Connect API integrations and set up your chat with MCPs.Start a chat to plan and work on a task.Not sure what to do? Just ask what seems the most important!
The more tightly scoped the task, the better cto.new will perform.Start with some small issues in your backlog to get some quick results and learn how cto.new works.You’ll son get used to working with cto.new and planning bigger tasks.
We recommend that you review cto.new’s code before merging it. This not only helps ensure it is correct, but keeps you in the loop on your own codebase.This step can be faster if you have test suites, lint checks, etc.It’s easiest to stay on top of cto.new’s work by reviewing PRs soon after they are made. This avoids code review piling up or PRs becoming outdated as your workflow speeds up.

​
Prompting tips

cto.new will usually do its best to complete a task but it may not be what you had in mind. Here are a few things you can do to make sure you’re both on the same page. Keep it short and to the point Try to exclude irrelevant context. Pasting in long AI generated task descriptions often doesn’t help. Be specific If there are specific files or pieces of code you want to edit, mention them! If there’s a data source you want to work with, add that too. If similar patterns or templates exist for a given task, it can help to refer to them. Tests? Let cto.new know what your testing requirements are for a particular task, or any other success criteria you have. Avoid vague tasks cto.new will attempt to complete vague or poorly defined tasks, but the outcome is unlikely to be what you had in mind.
⌘I