Once you're happy with a plan, executing it turns it into actual work. Unstoppable Code gives you three ways to do that, ranging from fully automated to fully manual. This article covers the three exits and when to use each.
Send it to a pipeline
Send the plan to a pipeline built to take a plan as input — a plan-ready pipeline. This runs the full implementation workflow in the background: refinement, frontend/backend partitioning, and QA loops, up through opening a pull request if it's set up to. It's the right choice for significant changes, and the usual one when you've spent time planning.
Just do it
Have the agent implement the plan directly. It works from the plan in a chat and gets started right away. This is a good fit for smaller, contained changes where you don't need the full pipeline machinery.
Copy the plan
Copy the plan's markdown and use it however you want — hand it to another tool, drop it into a different workflow, or keep it for reference.
Which to choose
- A significant feature, or a lot of work → send it to a pipeline. You get the full lifecycle, running in the background while you do other things. This is the default for anything you took the time to plan.
- A smaller, contained change → just do it.
- You want the plan somewhere else → copy the markdown.
Next steps
- Working in parallel — fire off a plan and move on to other work.
- Reading a pipeline execution — follow the run after you send a plan to a pipeline.
- Planning mode & plan tracking — where plans come from.