What problem does it solve? Technical interview knowledge scattered across websites and AI chat histories is hard to reuse, deduplicate, and keep accurate over time. This Skill retrieves, normalizes, verifies, and archives Java, backend, and computer-science interview questions into a maintainable Markdown knowledge base, and can also quiz you from that knowledge base. ## Core Features & Use Cases - Website Retrieval & Archiving: Browse trusted sources like JavaGuide, official docs, and specs, then archive questions with preserved heading hierarchy, source URLs, and stable IDs. - Conversation Distillation: Extract reusable interview knowledge from user questions and agent answers, with semantic dedup decisions (add, merge, revise, keep-conflict, migrate, or skip) shown in a coverage matrix before writing. - Knowledge Base Maintenance: Deduplicate, correct, split, merge, and restructure existing entries while maintaining root indexes and internal links, with optional Git commit/push. - Interview Quiz Mode: Draw questions from the saved knowledge base (with a configurable cooldown, default 14 days), ask in interviewer tone, score your spoken answer against compressed key points, and auto-archive practice records with commit+push. - Use Case: Say "从 JavaGuide 整理并发八股" to archive a Java concurrency topic, then later say "出一道八股" to be quizzed on a question you have not seen in two weeks. ## Quick Start Ask the agent to organize interview questions from JavaGuide about JVM into your knowledge base, or say "八股面试" to start a scored quiz session.