菜单
Administer a timed mock certification exam and score it on the 100-1000 scale (720 to pass). Use when the user wants a full practice run under exam-like conditions with a final score and per-domain breakdown.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/patrickking67/anthropic-university --skill mock-exam命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Author new original practice questions for an Anthropic University track (contributor tool). Use when working inside the anthropic-university repo to add questions to content/<exam>/questions.json.
Adversarially review an Anthropic University question bank for wrong keys, ambiguity, tells, weak distractors, and duplication (contributor tool). Use before opening a PR that changes questions.
Give a clear, docs-grounded explanation of a Claude Certification concept (e.g. prompt caching, the agentic loop, batch processing, .claude/rules, escalation design), with a worked example and a check question. Use when the user wants to understand a topic, not be quizzed.
基于 SOC 职业分类
正在显示 SKILL.md
| name | mock-exam |
| description | Administer a timed mock certification exam and score it on the 100-1000 scale (720 to pass). Use when the user wants a full practice run under exam-like conditions with a final score and per-domain breakdown. |
| argument-hint | <exam-id> [question-count] |
| allowed-tools | Read, Grep |
Administer a timed, exam-like run for $ARGUMENTS (exam id + optional question count).
associate-foundations, developer-foundations, architect-foundations,
architect-professional. Load ${CLAUDE_PLUGIN_ROOT}/data/<exam-id>.json (or, inside the repo,
content/<exam-id>/questions.json). Use its meta for the real format (question count, time,
passScaled).weight.meta.timeMinutes)flag, skip, review (list flagged), submitreview, list flagged/skipped ids
with one-line stems so they can jump back (re-ask that item).12/20 answered · 2 flagged), not
correctness.When finished (or the user says submit):
scaled = round(scaleMin + (correct/total) * (scaleMax - scaleMin)). Compare to
passScaled (720) → PASS / BELOW PASS.explanationCorrect, and why their
choice missed when applicable. Group by domain./exam-coach:quiz-me <exam-id> <weak-domain> or flashcards).Reminder: unofficial practice only. The scaled score helps calibrate study; it is not an official result.