Controlled Plan Execution FAQs

Question 1

When should I use this skill over subagent-driven development?

Accepted Answer

Use 'executing-plans' when you have a pre-defined plan file to follow. Use 'subagent-driven-development' for independent tasks or when working without a structured implementation roadmap.

Question 2

What is the difference between One-go and Batch mode?

Accepted Answer

One-go mode executes all plan tasks continuously and reports at completion, while Batch mode pauses after small sets of tasks (usually 3) to allow for human review and feedback.

Question 3

What happens if the automated code review finds an issue?

Accepted Answer

Critical, High, and Medium severity issues must be fixed immediately before the skill can proceed. Low and Cosmetic issues are documented with TODO and FIXME comments for later attention.

Question 4

How does it handle projects with multiple modules or directories?

Accepted Answer

The skill monitors the target directory for each task. When a switch is detected, it prompts for confirmation, changes the working directory, and loads any local PROJECT_RULES.md files.

Question 5

Can I skip the initial execution mode question?

Accepted Answer

No. The mode selection via AskUserQuestion is a mandatory structural gate. It ensures the correct architecture for review checkpoints and error recovery paths is established before work begins.

Controlled Plan Execution

Controlled Plan Execution

主要功能

使用场景

主要功能

使用场景