Tenstu
UserCrack algorithm coding interviews with LLM + MCP tools — a Claude Code harness for exam prep
Categories
Indexed Skills (2)
solve-analyze
Use when the user pastes their own solve() code and asks "分析一下我的代码", "帮我看看哪里错了", "对比一下", "review my solve", "check my code", "看看我的解法", "为什么WA", "为什么超时", "解法对比", "代码诊断". Also use when the user shares a completed solve() and wants a structured comparison against the standard template solution, root cause diagnosis, and automatic feedback to mistake_log / user_profile / exam-memory. This skill sits between solve-skeleton and algo-annotation in the pipeline: it diagnoses what went wrong, not how to annotate or how to build a skeleton. Handles both WA/TLE (full diagnosis) and AC code (lightweight style check).
solve-skeleton
Bare-bones Python solve() skeletons for AI Lab campus recruitment exam OJ problems. Use when the user asks for a solve() template, coding framework, OJ scaffold, or any of these Chinese trigger phrases: "解题框架", "解题模板", "solve骨架", "OJ模板", "ACM框架", "编程题结构", "solve()模板", "输入输出框架", "算法骨架", "解题规范", "代码模板", "框架代码", "答题模板", "刷题框架", "算法框架". Also use before starting any OJ problem implementation to provide the correct skeleton for the algorithm type. After the user fills in the TODOs, direct them to the `algo-annotation` skill for adding Chinese line-level comments. This skill provides structure only — no logic, no comments. For complete solutions, use `algorithms/python_oj_template.py` directly instead.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.