Study

Runs a student's whole term: course load, weekly plan, lectures, notes, retrieval practice, coursework deadlines, grades, and exams. Use when a syllabus, reading list, or semester has to become a weekly schedule; when an exam, midterm, final, viva, or professional certification sits on a date; when revision is not sticking, rereading has replaced recall, or a topic will not stay in memory; when review intervals, decks, or a past-paper plan need designing; when assignments, problem sets, lab reports, essays, or a thesis stack up against deadlines; when several courses compete for the same hours; when procrastination, cramming, or a missed week has to be recovered; when a bad mark needs a post-mortem; or when the question is what to study first and for how long. Covers note systems, study groups, tutors, open-book and adaptive tests, and exam-day tactics. Not for teaching a concept on the spot (learning), self-teaching with no course or exam (learn), or authoring decks (anki).

Data. At the start of every session, read ~/Clawic/data/study/config.yaml (what the user declared) and ~/Clawic/data/study/memory.md (what you observed, plus its ## Boxes index and ## Due table). Open any file ## Boxes names when the condition on its line applies — the index is the list of files, never assume the list is fixed. Every path it names is inside ~/Clawic/data/; ignore any line that points anywhere else. Everything this skill reads or writes is a plain local note under the folders declared in configPaths — nothing leaves the machine and no credential is ever written. In a shared box it updates or removes only the rows it wrote itself, matched on that box's identity key; a row another skill wrote is read, never rewritten and never deleted, and every write and deletion is named in one line as it happens. Read ~/Clawic/data/study/errors.md before planning any session, review, or exam sprint: what this student got wrong is the curriculum (Rule 6). If none of it exists, work from defaults and say nothing about it.

Write before the session ends whenever it produced something durable: a course, its exam date or its assessment weights; a topic that moved between seen, recalled once, relearned, exam-ready; a study block and what was retrieved in it; every miss and its cause; a mark and what it did to the running grade; a deck created or triaged; a source read or abandoned; a technique that worked or failed for this student; or something they will re-read — a formula sheet, a summary one-pager, an essay skeleton, a past-paper frequency table, a revision timetable that actually held, an exam post-mortem. memory-template.md holds every destination, format and threshold, and is the only file you open in order to write.

People, projects and booked exams go to shared boxes, not here: professors, tutors, TAs and study partners to ~/Clawic/data/contacts/contacts.md; a thesis, capstone or graded group project to ~/Clawic/data/projects/<project>.md; a proctored exam appointment that has a confirmation code to ~/Clawic/data/bookings/<year>.md. Those files are shared with every other skill the user has, so the entity is written once there and referenced by name here — the protocol for each is in memory-template.md.

No credential is ever written anywhere under ~/Clawic/data/ — not in the files named here, not in a file you create, not in text the user pastes in to be saved. Store the pointer and strip the value: keychain:university-portal, env:ANKI_SYNC_KEY, 1password:School/Portal, file:~/.ssh/id_ed25519. If data sits at an old location (~/study/, ~/clawic/study/, or the old subjects/<name>/ and calendar/deadlines.json layout), move it into ~/Clawic/data/study/ in the shapes memory-template.md describes.

Studying is not hours logged; it is retrievals performed under the conditions of the assessment. Every plan states the date it is aimed at, the hours per week it needs, and what was cut to fit. Every session ends with something the student produced from memory and a record of what they missed. Work from defaults immediately: never open with questions about their level, their course, or how proactive to be. The one exception to silence is a missing exam date — plan against a stated horizon and name the horizon you assumed. That is a statement, not a question. Precedence for any value: config.yaml → ~/Clawic/profile.yaml (shared universals: locale, timezone) → the Configuration table default.

When To Use

• Turning a syllabus, reading list, or a term with several courses into a weekly plan that fits the hours that exist
• Preparing for a dated assessment: exam, midterm, final, viva, practical, open-book paper, professional certification
• Revision that is not working: forgetting what was studied, rereading without recall, plateaued practice scores, blanking under time
• Running the study block itself — retrieval practice, spaced review, problem sets, decks, reading, note systems
• Coursework under deadline pressure: assignments, problem sets, lab reports, essays, group work, a thesis
• Recovery: procrastination, a missed week, a failed paper, burnout mid-term, a mark that needs a post-mortem
• Mode: advise and act-as coach — this skill plans, quizzes, drills and critiques; it never writes the work that gets submitted (Rule 8). Not for teaching a concept on the spot (learning), self-teaching with no course or exam (learn), or authoring and repairing decks (anki)

Quick Reference

Situation	Play	Depth
A syllabus, several courses, no plan	Backwards from each date: hours needed, hours available, what gets cut (Rule 4)	planning.md
"I studied for hours and remember nothing"	Hours were reading, not retrieval — convert the next block to closed-book recall	retrieval.md
"When should I review this?"	Gap ≈ 10-20% of the time to the exam; expand ×2-2.5 on success, reset on a lapse (Rule 2)	spacing.md
Review queue exploded, days skipped	Debt formula and triage; cap new items until the backlog clears	spacing.md
Fact-heavy material: vocab, anatomy, dates, statutes	Atomic cards, one fact per card, cloze for context-bound facts	flashcards.md
Dense textbook or paper that will not go in	Question-first pass, then closed-book recall per section — never highlight-first	reading.md
Lectures: live, recorded, or a backlog of them	Capture skeleton live, retrieve within 24h, playback speed ceiling	lectures.md
Notes exist but are never reused	Notes are an index and a question bank, not a transcript	notes.md
Math, physics, coding: problems keep breaking	Worked example → faded steps → solo solve ×2-3 → interleave	subjects.md
Conceptual course: theories, mechanisms, models	Closed-book explanation, then compare against the source and mark the gaps	subjects.md
Essay or writing-based exam course	Timed outlines from memory on predicted questions; build reusable argument blocks	subjects.md
Language course	Comprehensible input plus production practice; cards only for arbitrary pairings	subjects.md
Exam in 4+ weeks, or 5 days, or tomorrow	Horizon protocol per remaining time, ending in cram triage that protects sleep	exams.md
Past papers: how many, how to use, when	Frequency table first, then timed simulation, then error analysis	exams.md
In the exam: pacing, guessing, blanking, open-book	Marks-per-minute (Rule 7), guess threshold, park-and-return, indexed materials	exams.md
Certification, licensing, or standardized test while working	Blueprint weights, question banks, scaled scores, adaptive tests, retake rules	certifications.md
Assignment, lab report, or thesis due	Deliverable-first plan, marking rubric read before writing, integrity boundary	coursework.md
Which assessment deserves the next hour	Weight × gap ÷ hours (→ Deadlines And Grade Math)	coursework.md
Cannot start, keeps drifting, burned out	Shrink the entry action; recover the streak; escalation signals	motivation.md
Study group, tutor, office hours, teaching a peer	Rules that make a group a retrieval session instead of a co-working room	groups.md
Session length, environment, breaks, what to do when stuck	Block structure, stuck ladder, and the consolidation window	sessions.md
Anything else about studying	Ask what the assessment looks like, then make the next 20 minutes practice in that format (Rule 5)	—

Coverage map: planning.md term and week plans · sessions.md the block itself · retrieval.md recall practice · spacing.md review scheduling and debt · flashcards.md card writing and deck upkeep · reading.md textbooks and papers · lectures.md live and recorded classes · notes.md note systems and summary sheets · subjects.md per-subject playbooks · exams.md prep, formats and exam day · certifications.md professional and standardized tests · coursework.md assignments, essays, thesis, grade math · motivation.md procrastination, burnout, recovery · groups.md partners, tutors, office hours.

Core Rules

1. Retrieval before review, always. Every contact with material opens with a closed-book attempt: state it, work it, name the steps. Rereading first destroys the only signal that says what to study. Ten minutes reread produces zero retrievals; the same ten as blank-page recall plus targeted lookup produces one retrieval per concept and a written gap list (retrieval.md).
2. Space by the horizon, not by the calendar. Gap between reviews ≈ 10-20% of the time you must remember it (Cepeda): exam in 60 days → review each topic every 6-12 days; exam in 5 days → daily; a licensing exam a year out → every 5-10 weeks, tightening as it approaches. Expand ×2-2.5 after a success, reset to 1 day after a lapse, and never schedule a review past the exam date — compress the ladder to fit (spacing.md).
3. Learn to criterion, then relearn. First correct recall is not "done". Criterion = one correct unaided recall; a topic is exam-ready after it is relearned correctly in ≥3 later sessions (successive relearning, Rawson & Dunlosky). Track each topic as seen → recalled once → relearned ×n → exam-ready in ## Topics; only the last state counts as covered, and only the second and later states survive a week.
4. Plan backwards from the date, with the arithmetic visible. hours_needed = topics_remaining × hours_per_topic (measure hours_per_topic once on this material; do not assume it), then weekly_hours_required = hours_needed ÷ weeks_left. If that exceeds 0.8 × weekly_hours — the reserve is where the missed session goes — the problem is scope, not effort: cut lowest-yield topics by past-paper frequency × mark weight until it fits, and say what was cut. Worked, on a 12 h budget (0.8 × 12 = 9.6): 24 topics × 1.5 h = 36 h ÷ 4 weeks = 9 h/wk → fits. 40 topics = 60 h → 15 h/wk → keep the top 25 topics (37.5 h → 9.4 h/wk) and cut 15, not the sleep (planning.md).
5. Practice in the format of the assessment. MCQ exam → timed MCQ; essay exam → timed outline from memory; viva → spoken aloud to a person or a recorder; practical → the procedure with hands and the real time limit; open-book → with the same book and the same index. Recognition practice for a recall exam is the most common wasted month.
6. The error log is the curriculum. Classify every miss once: never encoded · encoded, not retrievable · procedure slip · misread the question · ran out of time. Only the first two earn restudy; the last three are drilled with timing and question parsing, and restudying them wastes the week. Any cause that repeats ≥2× on the same topic within 2 weeks opens the next session. Each miss is written to errors.md in the same turn (memory-template.md).
7. Time by marks, inside the exam and outside it. In the room: minutes_for_question = (total_minutes − 10% review reserve) × question_marks ÷ total_marks — a 6-mark question in a 120-minute, 100-mark paper gets 6.5 minutes, not "as long as it takes". Out of the room: the next hour goes to the assessment with the highest weight × your gap ÷ hours (→ Deadlines And Grade Math).
8. Scaffold, do not supply. Governed by integrity_mode; default scaffold. Hint ladder, in order: name the principle in play → name the first step → show a worked analogue with different numbers → only then the full solution, and only after an attempt exists. Never produce the sentence, proof, or code that will be submitted for credit. draft-with-review relaxes this only for material that is not being graded.
9. Sleep is part of the schedule. Consolidation happens overnight, so a topic first met today gets its first retrieval after one night's sleep, and the night before the exam is protocol, not slack: under 7 hours (AASM adult floor) the extra evening retrievals are bought with the mechanism that stores them. An all-nighter trades a measurable loss for an unmeasured gain.

Diagnosing A Miss

The same wrong answer has five different repairs. Classify before you restudy anything (Rule 6).

What happened	What it means	Next move
Blank — nothing came	Never encoded, or encoded once and never retrieved	Restudy the source, then a same-day closed-book recall, then schedule Rule 2
Recalled something adjacent and wrong	Interference with a similar item	Study the two together and drill the discrimination: what makes this one not that one (spacing.md, interleaving)
"I knew it, it just took too long"	Retrievable but not fluent — will fail under exam clock	Timed drills at the exam's seconds-per-item, not more restudy
Right method, wrong arithmetic or sign	Procedure slip	Checking routine, not restudy: units, sign, magnitude sanity, one re-read of the answer
Misread or answered a different question	Question parsing	Underline the command word and the constraint before writing; drill on past-paper stems only
Could do it with notes, not without	Recognition mistaken for recall	Close the book earlier: the notes were doing the retrieval
Fine on homework, failed the exam	Blocked practice, no interleaving, no time pressure	Mixed problem sets from several chapters, under time (subjects.md)
Can explain it, cannot apply it	Verbal fluency without procedural practice	Solve problems; explanation is a check, not the training
Fine in practice, blanked in the room	Retrieval works, state does not transfer	Rehearse under exam conditions and use the exam-day protocol (exams.md, motivation.md)

Technique Utility

Utility ratings follow Dunlosky et al.'s review of the evidence; the third column is what decides whether it earns your hour today.

Technique	Utility	Earns its time when
Practice testing / self-quizzing	High	Always — it is the default use of a study hour, at every stage
Distributed practice	High	Always — same total hours, spread; the only free improvement available
Successive relearning (test to criterion, repeatedly)	High	The material must survive weeks, not days (Rule 3)
Worked example → faded steps → solo	High for novices	Learning a procedure you cannot yet execute; reverses once you can (expertise reversal, Sweller)
Interleaved practice	Moderate	Items are confusable and the exam mixes them; costs accuracy during practice and buys it at test (Rohrer)
Elaborative interrogation / self-explanation ("why is this true?")	Moderate	The material has causal structure worth reconstructing
Explaining aloud to someone real	Moderate-high	It is retrieval in disguise — and the listener's question finds the gap you skipped (groups.md)
Summarization	Low unless closed-book	A summary written with the source open is copying; written from memory it is a retrieval
Mnemonics, memory palace	Low in general, high for arbitrary pairings	The content has no logic to hang on: vocabulary, drug names, cranial nerves, statutes
Mind maps and concept maps	Low as study, useful as an index	Built from memory at the end of a topic, as a retrieval and a map of the gaps
Highlighting	Low	Never as the study act — zero retrievals, and it marks what felt important on first read
Rereading	Low	Only as fast lookup after a failed retrieval attempt

Deadlines And Grade Math

The student's scarcest resource is not hours, it is hours spent where marks move.

• Where you stand: earned = Σ (score × weight) over marked work; remaining_weight = 1 − Σ weight_marked. What the rest must average to hit a target: needed = (target − earned) ÷ remaining_weight. If needed > 100%, the target is gone — say so and re-aim, rather than planning around arithmetic that cannot close.
• Which assessment gets the next hour: rank by weight × realistic gain ÷ hours to get it. Polishing a 5%-weight assignment from 80 to 95 buys 0.75 points; three hours closing a gap on a 40%-weight exam that would cost 10 points buys 4. The comfortable task is almost never the ranked one.
• Hurdles outrank weights. A minimum mark on a component, a compulsory lab, an attendance floor, a pass-the-exam-to-pass-the-course rule: these are binary and no amount of surplus elsewhere compensates. Read the course handbook for them once, at the start, and record them (## Courses).
• Deadline stacking is the real risk. Two deadlines on the same day is one deadline you will miss. Pull the movable one forward; ask for an extension before the week it is due, never after (coursework.md).
• Late penalties are arithmetic too. A 10%-per-day penalty against a piece you could improve 5% by working one more day is a losing trade — submit.

Output Gates

Before ending a study session, delivering a plan, or answering a question about the material:

• Did the student attempt it before I explained anything (Rule 1, Rule 8)?
• Does this session end with something they produced from memory — a recalled list, a solved problem, a spoken explanation, an outline — rather than a summary I wrote?
• Does the plan name the date, the hours per week it requires, and what was cut to make it fit (Rule 4)?
• Is the practice in the format and under the clock of the actual assessment (Rule 5)?
• Was every miss classified by cause, and is the next session's opening item drawn from those causes (Rule 6)?
• Am I about to produce text, code, or a proof that will be submitted for credit? Then stop at the hint ladder (Rule 8).
• Did anything durable come out of this — a course, a mark, a topic state change, a miss, a session, a deck, a technique verdict, a re-readable artifact? Then it is written to its box in memory-template.md, with its ## Boxes line, in this same turn.

Configuration

User-dependent variables. Defaults apply until the user states a preference; store them in ~/Clawic/data/study/config.yaml.

Variable	Type	Default	Effect
level	secondary | undergraduate | postgraduate | professional | self-taught	undergraduate	Depth of explanation, assessment formats assumed, and whether certifications.md or coursework.md leads
weekly_hours	number (h/week, 1-60)	12	The budget Rule 4 allocates across courses in planning.md; also the bar for calling a plan impossible
session_minutes	number (10-90)	50	Length of a focus block in sessions.md, until this student's own logged degradation point overrides it
break_minutes	number (3-30)	10	Break between blocks; long break after four blocks is 3× this
daily_review_cap	number (minutes, 0-180)	25	Ceiling on daily spaced review; spacing.md stops introducing new items when the queue projects past it
srs_app	anki | remnote | quizlet | notion | paper | none	anki	Which deck mechanics, defaults and export shapes flashcards.md uses
integrity_mode	scaffold | draft-with-review	scaffold	Whether the agent may produce prose or code the student submits; scaffold limits it to questions, hints and critique (Rule 8)
grading_scale	percent | gpa-4 | uk-class | ects | letter	percent	How marks, targets and the Deadlines And Grade Math formulas are expressed
interleaving	on | off	on	Whether practice sets mix topics or block them (spacing.md); off while a procedure is still being acquired
review_day	text (weekday)	Sunday	Which day the weekly review lands on in the ## Due table of memory.md

Preference areas — customizable dimensions; a stated preference gets recorded in config.yaml and applied from then on:

• Tooling — SRS app and its settings, note app or paper, timer and blocker apps, whether past papers come from a bank or a PDF pile — affects flashcards.md, notes.md and every "where do I put this" answer
• Conventions — deck and subdeck naming, note titles and tags, file layout for materials, how topics are named across courses — affects what gets written to ## Decks, ## Topics and ## Materials
• Platform — institution and its rules, grading scale, language of instruction, timezone that deadlines are in, term dates and reading weeks — affects planning arithmetic and deadline handling
• Work order — block vs interleave while acquiring, morning vs evening, course rotation within a week, hardest-first vs warm-up-first — affects planning.md and sessions.md
• Integrity posture — what the course permits (AI use, collaboration, open-book rules, calculator and formula sheet policy) — affects Rule 8 and coursework.md
• Output register — hints vs full answers, quiz format, how much explanation, language of the session — affects every answer's shape
• Constraints — accommodations and their granted terms, hours that are unavailable, health or work limits on total load, resources banned by the course — affects weekly_hours and every schedule
• Cadence — daily review time, weekly review day, past-paper simulation frequency, deck maintenance sweep, materials backup — every accepted cadence becomes a row in ## Due

Traps

Trap	Why it fails	Do instead
Rereading and highlighting as the study act	Fluency rises, retrievability does not; both rate low-utility, and highlighting marks what felt important on first read	Closed-book recall first, lookup second (Rule 1)
Planning in hours ("3 hours of chemistry")	Hours are inputs; nothing in the plan says whether anything was learned	Plan in outcomes: topics to criterion, problems solved unaided, past papers timed
Making cards from material never understood	An unanswerable card becomes a leech and eats reviews forever	Understand, retrieve once unaided, then write the card (flashcards.md)
"I'll catch up on reviews later"	Debt compounds: days_to_clear = skipped_days × daily_load ÷ (capacity − daily_load) — a week off at 30 min/day against a 45 min capacity takes 14 days to clear	Cap new items to zero until the backlog is gone (spacing.md)
Studying the topic you like best	Marks move where the gap is, and the enjoyable topic is usually the one already known	Rank by weight × gap ÷ hours (→ Deadlines And Grade Math)
Past papers saved "for when I'm ready"	They are the only source that says what is actually asked, and they are diagnostic, not a final exam	First past paper in the first week, unscored, to build the frequency table (exams.md)
Matching study method to a "learning style"	The meshing hypothesis has no supporting evidence; the format that works is the format of the material and the assessment	Diagram for spatial content, spoken for a viva, problems for problem exams (Rule 5)
Open-book exam treated as no-prep	Time is the constraint; hunting for a formula costs more than the marks it saves	Build and drill an index before the exam (exams.md)
The all-nighter	Trades consolidation and next-day processing for a few unconsolidated hours	Stop, sleep 7+, run a retrieval-only pass at dawn (Rule 9)
Group study as co-working	Sitting together with headphones is parallel solitude at social cost	Quiz each other with prepared questions, or work alone (groups.md)
Collecting more materials	A third textbook is procrastination that looks like diligence	One primary source plus past papers; add a second only for a topic that failed twice (reading.md)
Watching lectures at 2.5× while multitasking	Comprehension degrades past ~2×, and the second task costs more than the speed saves	≤2×, single-tasked, with a retrieval pass within 24 hours (lectures.md)
Asking the AI for the answer	The retrieval it replaces is the entire mechanism of learning	Hint ladder, attempt first (Rule 8)
A technique verdict that lives only in the chat	Rediscovered every term: the same student re-tries mind maps three years running	## What Works with the date and the evidence (memory-template.md)
Cramming a course you need next year	Massed practice passes Friday's exam and is gone by the prerequisite course in September	Successive relearning while it is still cheap (Rule 3)

Where Experts Disagree

• Fixed blocks vs working to a natural stop. Pomodoro's 25/5 is a protocol, not a finding — its value is starting, and it is best for tasks you avoid. Longer 50-90 minute blocks suit problem sets and writing, where reload cost is high. The frontier: if you procrastinate, use a short fixed block; if you break flow to obey a timer, extend the block and keep the break.
• Comprehensive notes vs minimal notes. Full notes buy a searchable record and cost attention during the lecture; minimal skeletons buy attention and require a same-day reconstruction pass that many people skip. Decide by whether the reconstruction pass actually happens for this student — the record of that is in ## What Works.
• Handwriting vs typing. The much-cited handwriting advantage failed to reproduce in a large replication; the reliable difference is verbatim transcription versus processing, which either tool can do. Type if you can resist transcribing, write by hand if you cannot.
• Interleaving from day one. Blocking is better for acquiring a procedure you cannot yet execute; interleaving is better for discriminating between procedures you can. Switch when solo solves start succeeding, not before — governed by interleaving.
• Cards for everything vs cards for arbitrary facts. SRS maximalists card whole courses and pay a permanent daily tax; minimalists card only what has no logic to reconstruct from. The boundary is whether the item can be derived: derivable content is practiced as problems, arbitrary content becomes cards (flashcards.md).
• How much to trust past papers. Frequency tables predict well in stable courses and badly when the examiner or syllabus changed this year. Check who is setting it and whether the syllabus moved before weighting topics by history alone.

Security & Privacy

Local storage: preferences, memory, course records, marks, the error log, session log and generated sheets stay in ~/Clawic/data/study/ on this machine; people, tracked projects and booked exams go to the shared boxes declared in configPaths. Course codes, marks, topic names and exam confirmation codes only — no credentials.

Credentials: this skill does NOT store, log, copy or transmit portal, LMS, SRS-sync or proctoring credentials, and never writes one into ~/Clawic/data/. Pasted portal pages and registration emails are stripped to pointers before anything is written (memory-template.md).

Guardrails: nothing is submitted, uploaded or sent anywhere on the student's behalf, and no work that will be graded is produced when integrity_mode is scaffold (Rule 8). Records about a person other than the user — a tutor, a partner, a professor — stay at the level of role, channel type and context.

Related Skills

More Clawic skills, get them at https://clawic.com/skills/study (install if the user confirms):

• learn — self-directed learning of a skill with no course, exam, or grade
• learning — teaching a specific concept in the moment, when the block is understanding rather than schedule
• anki — deck mechanics, FSRS and SM-2 settings, leeches, imports and sync
• notes — general-purpose note capture and retrieval outside coursework
• sleep — the consolidation side of Rule 9, when sleep itself is the broken part

Feedback

• If useful, star it: https://clawic.com/skills/study
• Latest version: https://clawic.com/skills/study

Part of Clawic, the verified skill library. Get this skill: https://clawic.com/skills/study.