DevKit Task Builder FAQs

Question 1

What is the difference between Sequential and Batch mode?

Accepted Answer

Sequential mode creates a separate Pull Request for every individual task and waits for a merge before continuing, while Batch mode executes all pending tasks on a single feature branch and creates one final PR.

Question 2

What happens if a dependency cycle is detected?

Accepted Answer

The skill performs a graph validation before starting. If a cycle is detected (e.g., Task A depends on B, and B depends on A), it will abort and list the conflicting tasks so you can resolve the logic in the tasks database.

Question 3

Can I filter which tasks are executed?

Accepted Answer

Absolutely. You can use --only to specify a list of IDs, --from to start from a specific order, or provide a text filter to match tasks associated with a specific parent prompt.

Question 4

Do I need to run the atomize skill before using build?

Accepted Answer

Yes, the build skill relies on a task database created by the atomize skill. If the database is missing, build will automatically attempt to invoke /devkit:atomize to generate the necessary task list.

Question 5

How does the skill handle task failures?

Accepted Answer

If a task fails during execution, the skill pauses and allows you to skip, retry, or abort. You can also use the --retry <id> flag to specifically reset and re-run a failed task later.

DevKit Task Builder

주요 기능

사용 사례

DevKit Task Builder

주요 기능

사용 사례