A developer has used Copilot Plan Mode to produce an agreed, ordered breakdown of the steps and files for a new feature. They now want Copilot to carry out that plan end to end, including making the edits and running the tests, with minimal step-by-step intervention. What is the appropriate next move?
- AStay in "Plan Mode" and instruct Copilot to begin applying each planned step directly, since the mode that produced the plan is best placed to execute it.
- BSwitch to "Edit Mode" and approve each individual edit one by one through the whole feature, which is the only way to apply a multi-step plan.
- CHand the agreed plan to "Agent Mode" so Copilot executes the steps autonomously, editing the files and running the tests with minimal intervention. Correct
- DPaste the plan into inline "ghost-text" completions in each file so Copilot fills in the implementation as the developer types through every step.
Why A is wrong: Plan Mode is for analysing and breaking down work before coding, not for autonomous execution, so it is tempting but does not carry out the edits and test runs.
Why B is wrong: Edit Mode requires per-edit review and does not run tests autonomously, which conflicts with the stated wish for minimal step-by-step intervention.
Why C is correct: Agent Mode is the surface built to execute multi-step work autonomously, so passing the Plan Mode breakdown to it matches the goal of carrying out the plan end to end.
Why D is wrong: Inline completions only suggest code at the cursor as a developer types and cannot orchestrate a multi-file plan or run tests, so they cannot execute the plan.