subagent-orchestration

You can orchestrate subagents by dispatching specialized implementers sequentially, using status-based decision logic to track task progress and handle blockers throughout the plan execution. This ensures end-to-end completion of complex development workflows.

Two-stage review runs a spec-review loop before transitioning to a code-quality review loop, iterating both stages until they pass. This staged approach validates implementation requirements first, then enforces code standards before marking the task complete.

Yes, the orchestrator supports per-task model tier selection when dispatching implementer subagents. You can assign appropriate model capabilities to each individual task within the plan, optimizing for complexity and resource allocation.

Task tracking uses status-based decision logic to monitor each dispatched subagent, automatically identifying blockers and orchestrating post-task verification. This provides continuous visibility into plan execution and ensures reliable end-to-end completion.

No, this orchestration handles ad-hoc plan execution outside the full triforce ceremony. You only need to provide a plan file to trigger the orchestrator, which manages subagent dispatch, iterative reviews, and post-task verification independently.

Staged subagent reviews are unnecessary for simple, single-step tasks that do not require spec validation or code-quality checks. If your workflow lacks complex multi-task plans or iterative review loops, lightweight execution is more efficient.