bd-execute

Plan-driven issue execution automates repository change workflows by claiming the issue, reading the referenced plan, and driving an implementer then spec-reviewer then code-quality-reviewer loop until all tasks pass.

Spec validation uses agent orchestration by injecting matching agent-rule digests based on the touch-set, then dispatching implementer and strict spec-reviewer agents that gate task progression until passing.

Enforce strict reviewer gating by running a verified dev loop where a spec reviewer and code-quality reviewer must both pass per task before final handoff, preventing unreviewed changes from proceeding.

Yes, issue execution optionally supports per-task model overrides using prompt templating, requiring superpowers template reads at runtime to drive the implementer and reviewer dispatches per task.

No, bd-execute does not directly call bd close; it forwards only to /bd-close-reviewed after all tasks are fully reviewed and code-quality-checked, ensuring strict operational control over the issue lifecycle.

If a bd issue has missing plan links or blocked conditions, the execution workflow handles them safely by performing claim and clean-state checks before proceeding with any implementation or review loops.