examprep-ai – skill Lập trình cho agent AI
Trợ lý ôn thi chuyển đề cương, đề thi cũ hoặc ghi chú thành Lộ trình Điểm Cao được xếp hạng. Bao gồm lý thuyết, bài tập tính toán, trắc nghiệm, lập trình và chuẩn bị thực hành, sắp xếp Dễ → Trung bình → Khó. Dùng để ôn tập phút chót, các chủ đề quan trọng và dự đoán câu hỏi.
Cài skill examprep-ai bằng một lệnh
Chọn công cụ bạn đang dùng, copy lệnh và dán vào terminal. Skill sẽ nằm trong thư mục skills của công cụ đó và sẵn sàng dùng ở phiên làm việc tiếp theo.
Claude Code
mkdir -p ~/.claude/skills && curl -fsSL https://skill.dn1nguoi.com/z/examprep-ai-8d50b.tgz | tar xz -C ~/.claude/skills
Gọi trong Claude Code: /examprep-ai – rồi mô tả công việc bạn muốn làm.
Codex
mkdir -p ~/.codex/skills && curl -fsSL https://skill.dn1nguoi.com/z/examprep-ai-8d50b.tgz | tar xz -C ~/.codex/skills
Gọi trong Codex: $examprep-ai – rồi mô tả công việc bạn muốn làm.
Hermes Agent
mkdir -p ~/.hermes/skills && curl -fsSL https://skill.dn1nguoi.com/z/examprep-ai-8d50b.tgz | tar xz -C ~/.hermes/skills
Gọi trong Hermes Agent: /examprep-ai – rồi mô tả công việc bạn muốn làm.
DeepSeek
mkdir -p ~/.deepcode/skills && curl -fsSL https://skill.dn1nguoi.com/z/examprep-ai-8d50b.tgz | tar xz -C ~/.deepcode/skills
Gọi trong DeepSeek: /examprep-ai – rồi mô tả công việc bạn muốn làm.
Nội dung SKILL.md
Nội dung SKILL.md gốc bằng tiếng Anh.
ExamPrep AI
When to Use
Use this skill when you need to:
- Convert a syllabus, past papers, or study notes into a prioritized roadmap.
- Focus on specific types of exam questions (Theory, Numerical, MCQ, Coding, Lab).
- Create flashcards, predicted exam papers, or check your overall exam readiness.
- Perform last-minute revision or deep-dive into important exam topics.
🎯 Selective Reading Rule — Read ONLY the section matching the request
| What the student asks for | Jump to |
|--------------------------|---------|
| Full roadmap / "what to study" / syllabus + past papers uploaded | [Full Roadmap Mode](#full-roadmap-mode) |
| Theory questions only / definitions / explanations | [Theory Notes](#theory-notes) |
| Numerical / calculation / derivation problems | [Numerical Notes](#numerical-notes) |
| MCQ / True-False / objective practice | [MCQ Notes](#mcq-notes) |
| Coding / algorithm / trace / debug | [Coding Notes](#coding-notes) |
| Lab / practical / viva prep | [Lab Notes](#lab-notes) |
| Flashcards only | [Flashcards](#flashcards) |
| Mock exam paper | [Predicted Exam Paper](#predicted-exam-paper) |
| Readiness check / score projection | [Exam Readiness Dashboard](#exam-readiness-dashboard) |
Rule: Read the matched section and the [Shared Foundations](#shared-foundations) block.
Skip everything else. Do not load all sections for a focused request.
---
Shared Foundations
> Load this block for every request. It is small and always needed.
Difficulty Scale (Universal)
| Level | Signal Words | Student Goal |
|-------|-------------|--------------|
| 🟩 Easy | define, state, list, name, identify, what is | Foundational recall — often useful first |
| 🟨 Medium | explain, describe, compare, calculate, implement, trace | Mid-paper marks |
| 🟥 Hard | derive, prove, optimize, analyze, evaluate, design, why | Score separators — study last |
Order rule: Always present Easy → Medium → Hard. Never reverse.
Intake (ask once, then proceed)
1. Collect at least one of: syllabus, past question papers, notes, or subject name + university.
2. Confirm course code when OCR is uncertain; do not invent a calibrated confidence percentage: *"I detected [X] — is this correct?"*
3. Ask time available. If no answer → default Standard Mode (6–12 hrs) and state the assumption.
Study Modes
| Mode | Time | Load |
|------|------|------|
| 🚨 Emergency | 1–2 hrs | 🟩 Easy only, top 10 questions |
| ⚡ Sprint | 3–5 hrs | 🟩 + 🟨, top 25 questions |
| 📚 Standard *(default)* | 6–12 hrs | All difficulties, full roadmap |
| 🗓️ Advance | Days+ | Daily schedule + mock papers |
Syllabus Guardrail
- Map every question to a specific syllabus unit using an actual topic match; label ambiguous mappings unconfirmed rather than inventing a 70% similarity threshold.
- Never generate content for topics absent from the uploaded syllabus.
- Out-of-syllabus items → flag, ask student before including.
Priority Score (not a probability)
Score = (Frequency × 0.40) + (Recency × 0.30) + (Unit Weight × 0.20) + (Marks × 0.10)- Frequency: appearances ÷ max appearances × 100
- Recency: last 2 yrs = 100 · 3–4 yrs = 60 · older = 30
- Unit Weight: core = 100 · elective = 50
- Marks: 10+ = 100 · 5–9 = 60 · 2–4 = 30 · MCQ = 20
Limitations
- This skill supports study planning and revision, but it cannot guarantee
exam questions, marks, grading outcomes, or instructor expectations.
- Priority scores are heuristics based on supplied syllabi, notes, and past
papers; sparse, outdated, or incomplete inputs reduce reliability.
- The skill should not fabricate syllabus coverage. If source material is
missing, ambiguous, or out of scope, ask the student to confirm before
adding predicted content.
- It is not a substitute for official course guidance, accessibility
accommodations, academic-integrity policies, or instructor feedback.
- Do not request or process private student records beyond the study material
needed for the current revision task.
---
Full Roadmap Mode
> Use when: student uploads syllabus + past papers, or asks "what should I study?"
Step 1 — Extract. Pull all questions; note year/source for each.
Confirm: *"Extracted [N] questions from [M] papers for [Course]. Found: 📝[A] 🔢[B] 🔘[C] 💻[D] 🧪[E]. Proceed?"*
Step 2 — Classify + tag difficulty. Use the five-type table:
| Type | Identify By |
|------|------------|
| 📝 Theory | define, explain, discuss, compare, differentiate |
| 🔢 Numerical | calculate, find, solve, derive, prove, numbers in question |
| 🔘 MCQ/T-F | options listed, "true or false", "which of the following" |
| 💻 Coding | write a program, implement, trace output, algorithm, flowchart |
| 🧪 Lab | experiment, procedure, observation, aim, apparatus, viva |
Step 3 — Build ranked tables (one per type):
| # | Question | Times | Marks | Difficulty | Unit | Priority |
|---|----------|-------|-------|------------|------|----------|
| 1 | [question text] | [N]× | [X] | 🟩/🟨/🟥 | Unit [X] | 🔥 Must / ✅ Do |Step 4 — Generate notes using the matching type section below.
Order: Easy across all types first → then Medium → then Hard.
Step 5 — Coverage tracker:
Unit 1: [Name] → 📝✅ 🔢✅ 🔘⚠️ PREDICTED 💻— 🧪—
Legend: ✅ past paper ⚠️ predicted — not applicableFor any gap: generate one predicted question + note, label [PREDICTED — not from past papers].
Step 6 — Offer: *"Would you like (a) Flashcards, (b) Predicted Exam Paper, or (c) Readiness Dashboard?"*
---
Theory Notes
> Use when: student asks about definitions, explanations, long-answer questions.
🟩 Easy — Definition / List (30 sec)
📝🟩 [Question] | [N]× | [X] marks
─────────────────────────────────
ANSWER: [2–4 bullets max]
KEY TERM: [single most important word]
MEMORY HOOK: [one-liner trick]🟨 Medium — Explanation / Comparison (2 min)
📝🟨 [Question] | [N]× | [X] marks
─────────────────────────────────
DEFINITION: [1 sentence]
MAIN POINTS: • P1 • P2 • P3 • P4
DIAGRAM: [text description — student sketches from this]
EXAM TIP: [what examiner rewards]🟥 Hard — Discussion / Evaluation (5 min read · 10 min write)
📝🟥 [Question] | [N]× | [X] marks | Unit [X]
─────────────────────────────────────────────
INTRO: [2–3 sentences]
SECTION 1 — [subtopic]: • point • point
SECTION 2 — [subtopic]: • point • point
SECTION 3 — [subtopic]: • point • point
DIAGRAM: [sketch description]
CONCLUSION: [1–2 lines]
MARKS HINT: Intro ~2 · each section ~3 · diagram ~2 · conclusion ~1
MEMORY: [acronym or order trick]---
Numerical Notes
> Use when: student asks for calculation problems, derivations, formulas.
🟩 Easy — Direct formula plug-in
🔢🟩 [Problem Type] | [N]× | [X] marks
──────────────────────────────────────
FORMULA: [clearly written]
GIVEN → FIND: [what's given / what to find]
WORKED EXAMPLE:
Step 1: [substitute]
Step 2: [calculate]
Answer: [result + unit]
COMMON MISTAKE: [the one error students make]
MEMORY HOOK: [how to remember formula]🟨 Medium — Multi-step with condition
🔢🟨 [Problem Type] | [N]× | [X] marks
──────────────────────────────────────
FORMULA(S): [all needed]
APPROACH: [which formula when — decision rule]
WORKED EXAMPLE:
Step 1: [setup / draw table]
Step 2: [apply condition]
Step 3: [calculate]
Step 4: [verify / interpret]
Answer: [result]
WATCH OUT: [condition that trips students]
EXAM TIP: [show working — marks for method too]🟥 Hard — Derivation / Proof
🔢🟥 [Problem / Derivation] | [N]× | [X] marks
───────────────────────────────────────────────
PREREQUISITES: [what student must know first]
DERIVATION:
Step 1: [first principles]
Step 2: [key transformation]
...Final: [result / QED]
WORKED EXAMPLE: [concrete numbers applied]
MARKS BREAKDOWN: [method marks vs answer marks]
COMMON ERRORS: [2–3 errors that lose marks]---
MCQ Notes
> Use when: student asks for MCQ practice, true/false, objective questions.
🟩 Easy — Recall
🔘🟩 [Question] | [N]×
──────────────────────
CORRECT: [option + text]
WHY CORRECT: [one sentence]
WHY OTHERS WRONG: • A: ... • B: ... • C: ...
KEY FACT: [the one thing this tests]🟨 Medium — Application
🔘🟨 [Question] | [N]×
──────────────────────
CORRECT: [option + text]
REASONING: [identify concept] → [apply rule] → [eliminate wrong]
TRAP: [why students pick the wrong answer]🟥 Hard — Trap / Edge-case
🔘🟥 [Question] | [N]×
──────────────────────
CORRECT: [option + text]
WHY TRICKY: [what assumption is exploited]
ELIMINATE: • Drop [A]: [reason] • Drop [B]: [reason] • Keep [C]: [reason]
RULE: [the precise rule that settles this type]---
Coding Notes
> Use when: student asks to write programs, trace output, implement algorithms, debug.
🟩 Easy — Syntax / Pattern recall
💻🟩 [Task] | [N]× | [X] marks
────────────────────────────────
PATTERN: [algorithm/structure name]
TEMPLATE: [minimal working skeleton — pseudo… (nội dung đầy đủ có trong gói tải về)
Skill liên quan
- executing-plans – Dùng khi bạn có kế hoạch triển khai bằng văn bản cần thực hiện trong một phiên riêng với…
- expo-api-routes – Hướng dẫn tạo API routes trong Expo Router với EAS Hosting
- expo-brownfield – Tích hợp Expo và React Native vào ứng dụng iOS hoặc Android native có sẵn. Dùng khi…
- expo-module – Hướng dẫn tạo và viết module và view native của Expo bằng Expo Modules API (Swift…
- zarr-python – Lưu trữ và truy vấn mảng khoa học N chiều phân khối (chunked) với Zarr-Python 3, bao gồm…
- faf-go – Phỏng vấn có hướng dẫn để đạt Gold Code (100% sẵn sàng cho AI). Dùng khi giúp người dùng…
- web3-testing – Kiểm thử smart contract toàn diện bằng Hardhat và Foundry với unit test, integration…
- roast – Dùng khi ai đó yêu cầu roast một ý tưởng, kiểm tra áp lực hoặc stress-test một ý tưởng…