Design an AI-Assisted Development Workflow

Explain how you use AI tools in software work and how you would integrate an AI agent into a practical development workflow. Use a concrete example, and state which decisions and checks remain outside the model. ### Part 1 — Describe everyday use Choose a task for which you use, or would use, an AI tool. Explain what you provide, what you expect it to produce, and how you check the result. #### What This Part Should Cover - A bounded task and relevant context. - Independent verification and limitations of the output. ### Part 2 — Design an AI workflow How would you organize the steps, tools, intermediate results, and failure handling for that task? #### What This Part Should Cover - Planning, execution, feedback, and stopping conditions. - Tool permissions, review boundaries, and measurements of usefulness. ### Part 3 — Direct an agent How would you prompt the agent and use reusable instructions, skills, or hooks where the chosen host supports them? #### What This Part Should Cover - Clear objectives, constraints, and evidence required for completion. - The distinction between model instructions and checks enforced by the surrounding application. ### What a Strong Answer Covers - A concrete workflow rather than a list of AI product names. - Controls that match the task's consequences and keep outputs reviewable. - Evidence of correctness and time saved, including the cost of review and rework. ### Follow-up Questions - What would you do if the agent repeatedly changed code without resolving the observed failure? - Which check should run as a deterministic host action instead of relying on the agent to remember it?