David Ondrej, Alex Lieberman and Dan Zakon treat context as part of the software system rather than a disposable promptContext as code stores important project intent, architecture, conventions and decisions in versioned files that AI agents and people can reuse.. Project goals, architecture, conventions and decisionsContext engineering designs the information, instructions, memory, and tool state an AI receives so it can perform a task reliably. live in versioned files so each agent starts from the same durable understandingAn AI coding agent is a tool-using AI system that can inspect, modify, and validate software within a repository..
David Ondrej, Alex Lieberman and Dan Zakon separate planning, implementation and reviewTask planning breaks an objective into ordered, bounded steps with dependencies, responsibilities and evidence for completion. while keeping tests and explicit completion criteriaAn acceptance criterion is a specific condition that must be satisfied before a software change or task can be considered complete. close to the work. This gives long-running agents useful checkpoints and makes human review focus on decisions instead of rediscovering intent.
David Ondrej, Alex Lieberman and Dan Zakon describe agentic engineering as an operating model for teams, not just faster code generation. Reliable results come from maintaining the context system, measuring failures and improving the reusable instructions around recurring work.
Watch on YouTube
