agentleFS
Cassius2584Cassius2584/claude-uni-workspace/.claude/skills/exam-practice/SKILL.md

exam-practice

Skill5 starsChanged 25 days agoMITView on GitHub

Exam-style practice for a module. Sets a past-paper or exam-style question (definitions, statements, short proofs, calculations) on results the student should know, has them answer by hand and send a photo, marks it strictly against a mark scheme written in advance, and logs scores and weak spots with next-review dates. Also runs timed mock papers. Use when the student says "quiz me", "give me an exam question", "test me on <topic/chapter>", "exam practice", "mark my answer", "mock exam", or sends a photo of a handwritten answer to a question you set.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read the module's MODULE.md, its assessment notes (exam format, length, closed or open book), the tiers page (proof-tiers.md or similar, …

    line 18
  2. Very specific:

    Read practice.md if it exists (create it from reference/practice-log.md if not, and add a link to it from the tiers page).

    line 21
  3. Distinctive:

    Read the photo carefully: If a line is unreadable, ask about that line rather than guessing, and never give credit for something you can't …

    line 51
  4. Distinctive:

    Don't give marks for effort, length, or things that weren't asked.

    line 60
  5. Very specific:

    If a definition or statement was wrong, offer to add or fix its flashcard (create-flashcards).

    line 76
  6. Distinctive:

    Practice only: Don't answer or mark anything that will be submitted for credit.

    line 85

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Exam practice

Practice that looks like the exam: a question with marks, answered by hand under exam conditions, photographed, and marked like an examiner would. Everything is recorded in the module folder:

projects/<module>/
├── practice.md                ← log of every attempt, weak spots with next-review dates, past-paper questions used
└── practice/.pending-scheme.md ← the mark scheme for the question currently set (hidden in Obsidian: dot-file)

0. Before you start

  • Read the module's MODULE.md, its assessment notes (exam format, length, closed or open book), the tiers page (proof-tiers.md or similar, from create-flashcards), flashcards/cards.md if present, the lecture notes, and the past papers and solutions in resources/.
  • Read practice.md if it exists (create it from reference/practice-log.md if not, and add a link to it from the tiers page).
  • Scope: during teaching, only chapters that have been lectured (MODULE.md weekly plan, the latest projects/updates/ digest, or ask). In the revision period, everything examinable. Skip anything the notes or the tiers page mark non-examinable.

1. Choose the question

In order of preference:

  1. Weak spots due for review (practice.md, next-review date today or earlier).
  2. A real past-paper question on lectured material that hasn't been used yet (practice.md lists the ones used). Only use the parts the paper marks as relevant to this year's module. Past papers are the best practice, but there are few of them, so save full papers for mock exams in the revision period.
  3. An exam-style question you write, modelled on the past papers' style and mark split: typically a definition or statement (2–5 marks), a short proof of a Tier A result or an unseen lemma using the same technique (4–8 marks), and a calculation (3–6 marks). Tier B results: ask for the key idea and outline, or one step. Rotate chapters, and favour results the student hasn't practised yet. The student can override: "test me on Euler's criterion", "a calculation question", "a whole past paper".

2. Set it

  1. First, write the mark scheme to practice/.pending-scheme.md: each part's marks, what earns each mark (precise definition wording, the key steps of the proof, method marks and accuracy marks for calculations), common errors to penalise, and a model answer. For a past-paper question, base it on the official solutions and their mark split. Don't show it, and tell the student it's there and not to open it.
  2. Then present the question in chat: numbered parts with marks in brackets [4], maths in LaTeX, and a suggested time (use the paper's pace if known, e.g. 120 minutes for 80 marks = 1.5 min per mark; otherwise 1.5 min per mark).
  3. Conditions: closed book if the exam is (no notes, no flashcards). Answer on paper, then send a photo of each page, in order. Typed answers are fine if they prefer.

3. Mark it

  • Read the photo carefully. If a line is unreadable, ask about that line rather than guessing, and never give credit for something you can't read.
  • Mark against the scheme, strictly and fairly, as a university examiner would:
    • Definitions and statements: full marks only for the exact meaning with every hypothesis ("gcd(a, m) = 1", "p an odd prime"). Missing a hypothesis loses a mark. Wrong or vague wording loses the mark.
    • Proofs: marks for each key step in the scheme. Correct alternative proofs get full credit. A proof that assumes what it's meant to show, or skips the key step, gets only the marks for the steps it actually shows.
    • Calculations: method marks for a correct method even when the arithmetic slips (follow-through), and accuracy marks only for correct values.
    • Don't give marks for effort, length, or things that weren't asked.
  • Feedback, in this order:
    1. A score per part and in total (14/20).
    2. For each lost mark: what was missing or wrong, quoting their line, and the correct version.
    3. One "remember this" line per weak point: the exact phrase or key step to learn.
    4. The model answer for any part scoring under half.
  • Then delete practice/.pending-scheme.md, and put the model answer in the log entry instead.

4. Log it

Add to practice.md:

  • a row in Attempts: date · question (source: "2025 Q2(c)" or "set") · results tested (links to tiers or cards) · score · one-line note;
  • for each weak point, a row in Weak spots: result · what went wrong · next review. Schedule the next review from the score on that result: under 50% → 2 days, 50–79% → 7 days, 80%+ → 21 days. A spot scoring 80%+ twice in a row moves to Cleared;
  • past-paper parts used, in Past papers used. If a definition or statement was wrong, offer to add or fix its flashcard (create-flashcards). Tiers are the lecturer's call, so don't change them. Do flag it if a Tier B result keeps coming up in questions.

5. Next

Offer another question: a weak spot, the next chapter, or "same result, different angle". In the revision period, suggest a timed mock: the relevant questions from one past paper, in one sitting, at the paper's pace. Mark each question as above, then give a total and the three weakest areas.

Rules

  • Practice only. Don't answer or mark anything that will be submitted for credit. For formative problem sheets, give hints and check reasoning (not full solutions), and only if the module's GenAI rules allow it.
  • Past papers and solutions are the university's copyright: quote what's needed for the question and the marking, keep them in the module folder, and never commit or share them.
  • Photos of answers stay in the conversation. Save them into practice/attempts/ only if the student asks.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.