Agile Scrum, Agentics, and the Skipjack Protocol: Governed Collaboration Between Human Judgment and Agentic Execution

Agile Scrum has governed software development execution for more than two decades. Its core commitments — iterative delivery, transparency, team collaboration, and responsiveness to change — remain valid in the age of AI. What Scrum was not designed for is a world in which artificial intelligence agents actively participate in execution: not as tools that human team members use, but as actors that decompose, execute, iterate, and report within the same workflow as their human counterparts. This paper bridges that gap. It extends traditional Agile Scrum into an agentic operating model — a framework in which human teams and AI systems function as a unified execution layer, governed by a shared protocol. That protocol is the Skipjack Protocol. The extension works by remap, not replacement. The Product Owner maps to a Governance Layer of humans with domain authority who define intent, constraints, and prioritization. The Scrum Master maps to an Orchestrator responsible for task routing, flow control, and system health. The Development Team maps to an Agentic Execution Layer of AI agents and human operators who execute within governed boundaries. The Backlog maps to persistent memory and task queues, maintaining epistemic continuity across sprint cycles. The Skipjack Protocol governs this system through four enforcement mechanisms: context integrity and epistemic continuity; structured task decomposition and routing; human-in-the-loop control points; and iterative feedback loops that close between human judgment and agentic execution. Together, they ensure that the three-tier architecture operates with integrity at any scale. Paper Five is the fifth paper in the Mission-Ready AI pentalogy. The key to scaling AI is not autonomy alone — it is governed collaboration between human judgment and agentic execution. Rights envelope: Citation permitted with full attribution. No reproduction, redistribution, or derivative works without written permission. AI/ML training use disallowed. See the citation policy at https://nonsequitur.tech/pubs/citation-policy/ for the full rights envelope. Canonical site URL: https://nonsequitur.tech/white-papers/skipjack-protocol/ Public archive: yks-pubs/papers/skipjack-protocol-v1-preprint.pdf