implement

Parallelize subagents for sprint validation by coordinating a deterministic run of three concurrent subagents against the Acceptance Contract. This enables faster convergence or timely escalation by processing per-cycle inputs in a single assistant turn.

Multi-agent orchestration for sprint convergence is the process of running Generator, Validator, and Orchestrator subagents simultaneously. It manages per-cycle inputs from sprint and progress state to drive convergence or escalation.

Manage canonical memory writes after a sprint cycle by applying post-iteration persistence and a final canonize handoff. The Orchestrator subagent invokes the canonization process to write converged sprint outcomes to canonical memory.

Yes, multi-agent orchestration works with per-cycle sensor tiering. The process drives sensor tiering alongside post-iteration persistence and the canonize handoff to validate sprint cycles against the Acceptance Contract.

You need to escalate sprint cycles instead of converging them when the parallel run of three subagents fails to satisfy preflight validation or the Acceptance Contract. This timely escalation prevents stalled sprint progress.