exam-curriculum-trainer

📁 panchok50/exam-curriculum-trainer 📅 5 days ago
1
总安装量
1
周安装量
#47877
全站排名
安装命令
npx skills add https://github.com/panchok50/exam-curriculum-trainer --skill exam-curriculum-trainer

Agent 安装分布

amp 1
opencode 1
kimi-cli 1
github-copilot 1
gemini-cli 1

Skill 文档

Exam Curriculum Trainer

Use this skill to create or update an exam-prep workflow in any project that contains course materials.

Announce And Scope

Start by stating you are using exam-curriculum-trainer and that the goal is exam performance, not broad theory coverage.

Collect only the minimum required inventory:

  • Curriculum or study-plan files already present.
  • Previous/mock exams and exercise sheets (highest priority sources).
  • Lecture notes/slides (secondary gap-filling source).

Prefer fast file discovery commands.

Priority Policy

Apply this weighting in all outputs:

  1. Previous exams and mock exams.
  2. Exercise/problem sheets and solved tasks.
  3. Lecture theory, only when needed to support exam-style solving.

Do not produce long theory digressions.

Create Curriculum File

Create a CURRICULUM.md (or update existing one) with the structure in references/curriculum_structure.md.

Hard requirements:

  • Keep ASCII only.
  • Include explicit training rules: no hints on first attempt, strict grading, weak-tag tracking, retake until 2 consecutive correct answers per weak tag.
  • Include a question bank mapped to source material IDs.
  • Include a fixed multi-session sequence and retake policy.
  • Include a session script with step-by-step runtime behavior.
  • Include concise formula sheet and pitfalls list.

If an existing curriculum already follows a strong structure, preserve it and patch gaps instead of rewriting blindly.

Build Exam-Style Question Bank

Extract and normalize questions from prioritized sources.

Rules for question writing:

  • Prefer exam-style prompts with explicit deliverables.
  • Keep each question atomic and gradable.
  • Add point values and weak-tag mapping.
  • Add near-variant follow-ups for remediation.
  • Include timed mixed sets after topical blocks.

When source questions are sparse, generate adjacent variants in the same style and notation.

Run Interactive Sessions

When user says to start a session:

  1. Load the configured session range from CURRICULUM.md.
  2. Ask one question at a time in strict mode.
  3. Grade each answer immediately: Correct/Partial/Wrong with points and short reason.
  4. If wrong/partial, provide corrected solution and one near-variant follow-up.
  5. Update weak tags inline and track streak-to-mastery.
  6. Continue automatically until session set completes or user pauses.

Session behavior requirements:

  • No hints on first attempt.
  • If user requests help, apply hint ladder (light -> medium -> strong).
  • Keep notation consistent with curriculum formulas.
  • Keep answers concise and terminal-readable ASCII.

Use references/session_runner.md for exact runtime protocol.

Output Standards

Use this response order when reporting session progress:

  1. Question result and score.
  2. Minimal correction.
  3. Weak-tag update.
  4. Next question.

At session end provide:

  • Total score.
  • Wrong/partial IDs.
  • Weak-tag status.
  • Required retake set.
  • 60-second recall prompt (3 formulas, 2 pitfalls, 1 anchor).

File Update Rules

When creating assets in a new project, generate only what is needed:

  • CURRICULUM.md mandatory.
  • EXAM_SIM_1.md and EXAM_SIM_2.md optional but recommended for timed full runs.
  • Do not create extra docs outside study workflow unless asked.

Adaptation Rules For New Courses

Keep structure stable but adapt content:

  • Rename topic blocks to match the course.
  • Preserve fixed sequence concept from fundamentals -> advanced -> timed exams.
  • Preserve mastery gating and retake loops.
  • Preserve exam-first source weighting.

If user gives a different emphasis, follow user preference while retaining strict grading and interactive practice defaults.