sihaoliu/vibe-agentic-interview-notes

This repository is an interview question bank, written for hiring developers in a world where AI coding assistants like Claude Code and OpenAI Codex are part of daily work. The author is upfront that it is one person's opinion about what interviews should look like in the AI era, and that there is no single correct answer for any of the questions. The reference answers attached are described as starting points for a conversation, not a rubric. The questions are aimed at engineers who already use agentic command-line tools as part of their workflow. The goal is to probe practical fluency and awareness of trade-offs, rather than rote recall of facts. Open issues and pull requests are welcomed for corrections or new questions. The bank is split into six sections. Background Knowledge covers foundational concepts like commands, skills, agents, MCP, and sub-agents. Detail Walkthrough goes into the mechanics of files like CLAUDE.md, plus hooks, permission modes, the /compact command, plugins, and ToolSearch. Workflow asks how to break down features, debug, and manage context inside a team. System Design poses open-ended prompts about low-code agentic platforms, MCP gateways, multi-agent topologies, and long-term memory. Concepts and Philosophy asks broader questions about agent autonomy and the future of programming. A final section called The Hidden Lore covers deeper details. The questions exist in both Chinese (the original, treated as authoritative) and English.