A reusable Claude Skill that turns your own course material into a two-in-one deliverable: a study guide and an evidence-ranked prediction of what the exam will actually ask, optimized for marks.
Built for biology-family courses — biotechnology, biology, biochemistry, bioinformatics, chemistry, microbiology, molecular biology, and related lab sciences — but every section is conditional, so it adapts to the material you give it.
- Study guide + exam prediction in one pass. Not a generic summary — it links theory → practical step → how it was asked before → expected answer.
- Evidence-weighted predictions. Every predicted question is tagged HIGH / MEDIUM / LOW confidence with the signals that justify it. Past exam papers are the strongest signal, then model answers, slide repetition, professor verbal emphasis, your insider tips, and textbook prominence last.
- Reads Arabic professor transcripts. Parses spoken Arabic (and code-switched Arabic + English) lecture transcripts, catches spoken exam cues like "ده مهم للامتحان" / "هييجي في الامتحان", and folds them into the English output as verbal-emphasis signal.
- Marks-driven prioritization. Asks for the total marks and distribution, then allocates depth and predicted questions proportionally.
- Conditional sections. Lab machinery (CFU/BOD/COD/enzyme-activity calculations, media/specimen identification tables, protocol breakdowns, spot-photo ID) activates only when the material justifies it.
- Source-faithful. Builds from your sources and preserves the professor's exact terminology. Outside knowledge is used only for central-but-poorly-explained concepts and is always visibly flagged.
- Two output formats. Self-contained color-coded HTML (default) or a printable Word
.docx.
Two ways:
- In the Claude app / claude.ai — download
dist/exam-study-guide-predictor.skilland use the Save skill button on the file card, or upload it in Settings → Capabilities → Skills (availability depends on your plan/org). - Claude Code / API — place the
exam-study-guide-predictor/folder in your skills directory.
Once installed it triggers on prompts like:
- "Help me study for my microbiology final"
- "Predict my biochem midterm from these slides"
- "Make me a revision guide" (with lecture files attached)
It then interviews you on: course, exam type, format, total marks + distribution, which sources you have, deliverable (guide / prediction / both), depth (deep / rapid / both), and output format.
Sources it can use: lecture slides, professor audio/voice-note transcripts (English or Arabic), past exam papers, official model answers, lab/practical manuals, textbook chapters, class notes, and your own "the Dr said…" insider tips.
exam-study-guide-predictor/
├── SKILL.md # workflow + intake interview
└── references/
├── prediction-and-analysis.md # signal weighting, Arabic cue tables, faithfulness rule
└── output-format.md # section catalog + HTML/.docx templates
dist/
└── exam-study-guide-predictor.skill # packaged, installable skill
Derived from a set of exam-prep prompts by Hossam Mahmoud, Abdelrahman El-Dbeky and Ahmed Hussein (Eddie), generalized from a single Environmental Biotechnology course into a reusable, discipline-adaptive skill.
MIT — see LICENSE.