backlog-implement

Safely implementing backlog items requires a verification-first workflow that checks current code reality, scores impact, and enforces a mandatory user confirmation gate before any code changes occur. This prevents wasted effort and risky edits.

Scope control for multi-file changes is enforced through staged workflow execution with mandatory approval gates. The implementer must verify the problem still exists, produce an impact score, and get explicit user confirmation before modifying code within the approved scope.

Yes, you can run tests automatically after implementing a backlog task. The implementation and verification loop performs the scoped code change, runs appropriate tests to verify the output, and then conditionally closes the backlog item upon success.

Verifying if a backlog task is still relevant involves a verification-first step that inspects relevant files, functions, and configuration state. The implementer agent checks whether the described problem still exists and records the findings before planning any changes.

To start implementing a specific backlog item, you can pass the backlog ID directly to the implementer agent. The workflow then loads the item context, verifies the current code reality, and prompts for user confirmation before proceeding with implementation.

No, you do not need to manually close backlog items after code verification. The workflow handles conditional backlog closing automatically once the scoped code changes are applied and the appropriate tests are run successfully.