Trunk-based with two agents
Your team practices trunk-based development on a Python service. You and a teammate are both using AI agents on overlapping areas of the codebase this sprint, integrating to main several times a day. Design the working agreement that lets two AI-assisted engineers commit fast to the same trunk without their agents' changes colliding or undoing each other. What's different about coordinating agent-driven work vs. two humans hand-coding?
find_trunk_collisions(edits: list[str], max_divergence_hours: int) → list[str][["alice|payments|1","bob|payments|9","alice|billing|2"],4]out["payments"][["alice|payments|1","bob|payments|3"],4]out[][["alice|payments|1","alice|payments|20","bob|search|2"],4]out[]Treat the AI’s output as a draft to verify, not an answer to trust. Name the specific flaw and the input that triggers it, say how you’d catch it (tests, edge cases, reading critically), and how you’d re-prompt or decompose to get it right.
Vibe & agentic: describe the solution in plain language (or narrate it) and the coach grades your approach.