universal-examprep-skill
Exam Cram Coach · 跨会话记忆与讲义溯源防幻觉的极速备考教练 | AI exam-prep tutor for Claude Code, Cursor, Codex, Antigravity: teaches from slides with page citat
What is universal-examprep-skill?
Exam Cram Coach · 跨会话记忆与讲义溯源防幻觉的极速备考教练 | AI exam-prep tutor for Claude Code, Cursor, Codex, Antigravity: teaches from slides with page citations, crops figures, quizzes with real homework, runs smoothly on small cheap models. 期末复习·考前突击·刷题·错题本·小抄
Overview of universal-examprep-skill
What it is
Exam Cram Coach (Flash edition) is an AI exam-prep tutor skill for coding agents such as Claude Code, Codex, Cursor, Windsurf, Antigravity, and Gemini CLI. It teaches from user-provided lecture slides, notes, homework, and past papers, with page citations, figure cropping, and quiz generation from real homework. It is designed to run on small, cheap models and maintains cross-session memory without external vector databases. The skill is approximately 1,600 words and consists of 13 files (about 150 KB). The tool runs offline.
Key features
- 100% grounded and anti-hallucination: every taught concept cites exact file and page number; out-of-scope queries exit with non-zero status; transparent 🟢/🟡/⚠️ provenance tags distinguish materials from AI commentary.
- Cross-session persistent memory: maintains state locally in
study_state.jsonand a mistake log; users can close and resume withcoach.py status; mistakes are prioritized in spaced re-quizzing. - High-end performance on cheap/weak models: deterministic CLI output with
📍 next commandhints; automated vector-figure cropping viapypdfium2; benchmarked on 1,000+ pages of real college probability notes with Gemini Flash-Lite and Claude Haiku achieving zero command fabrication and zero phantom citations. - Reads PDF, PPTX, DOCX, Markdown, TXT, HTML and splits the course into chapters.
- Cuts figures out of lecture notes and printed questions/solutions.
- Pulls real questions and reference answers from homework and exams (matches
hw2.pdfwithhw2solutions.pdf). - Teaches chapter by chapter, quizzes only with those questions, tracks mistakes, and compiles a concise cheat sheet.
Use cases
- Exam preparation and last-minute cramming for any course with lecture slides, notes, homework, and past papers.
- Studying with small or local AI models where token efficiency and low cost are important.
- Cross-session study: closing and resuming a study session without losing progress or mistake tracking.
- Generating a cheat sheet from course materials.
Getting started
bash npx skills add ZeKaiNie/universal-examprep-skill pip install pypdfium2 Requires Node.js 18 or newer for npx and Python 3.8 or newer. For alternative installation methods (Claude Code plugin, offline zip), see the repository README.
License and pricing
License: MIT. Pricing type: open source.
Based on the project's public README and repository metadata.