learn-lesson
/learn-lesson
Uczy jednego modułu kursu jako aktywną lekcję: hak, krótkie wyjaśnienie, przykład rozwiązany, przykład do dokończenia i trzy pytania przypominające.
- Autor
- vivekbhange-hub ↗
- Repozytorium
- learn-anything-skill ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 07.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 07.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Uczy jednego modułu kursu jako aktywną lekcję: hak, krótkie wyjaśnienie, przykład rozwiązany, przykład do dokończenia i trzy pytania przypominające.
Wyzwalacze z opisu
Argumenty
/learn-lesson [module id, e.g. M2]
Oryginalny opis (EN)
Teaches one module of a learn/<topic>/ course as an active lesson: a hook, a short explanation, a fully worked example, a faded example the learner finishes themselves, then three retrieval questions. Depth follows the learner's diagnosed level. Saves the lesson to learn/<topic>/lessons/. Use when the user says "teach me M2", "next lesson", "start module 1", "explain closures properly", "continue the course", "let's do today's lesson", or "I'm ready for the next part".
Jak działa
- Inputs
- Steps
- Decision rules
- What NOT to do
Podgląd SKILL.md
---
name: learn-lesson
description: >-
Teaches one module of a learn/<topic>/ course as an active lesson: a hook,
a short explanation, a fully worked example, a faded example the learner finishes
themselves, then three retrieval questions. Depth follows the learner's diagnosed
level. Saves the lesson to learn/<topic>/lessons/. Use when the user says "teach me
M2", "next lesson", "start module 1", "explain closures properly", "continue the
course", "let's do today's lesson", or "I'm ready for the next part".
argument-hint: "[module id, e.g. M2]"
---
# learn-lesson
Teaches one module the way learning research says works: explanation is short, the
learner studies a complete worked example, then does a partially worked one where
the final steps are removed (fading), then pulls the ideas back out of memory with
retrieval questions. Reading a good explanation feels like learning; producing the
answer is what actually builds it. This skill makes sure the learner produces.
## Inputs
- `learn/<slug>/syllabus.md` and `learn/<slug>/learner.md` (from learn-syllabus).
If there are several course folders under `learn/`, ask which one. If none exists,
offer to run **learn-syllabus** first, or teach a one-off lesson without saving.
- Module id from the argument. With no argument, pick the first module with no file
in `lessons/` whose prerequisites all have lessons.
- `learn/<slug>/tours/NN-*.md` for a codebase course built by **learn-repo**: the
tour is the source for that module, and its stops are where examples come from.
- `learn/<slug>/mistakes.md` if present: open mistakes in this module's prerequisites
are worth one sentence of repair in the hook.
## Steps
1. **Read the module** in syllabus.md: objectives, prerequisites, time. Read
learner.md for level and preferences. If a prerequisite module has never been
taught, say so and offer to do it first; continue only if the learner insists.
2. **Set depth by level.**
- `new`/`beginner`: concrete before abstract, one idea per step, everyday analogy,
define every term on first use, small numbers and short code.
- `intermediate`: lead with the mental model, contrast with what they already know,
include one common misconception.
- `advanced`: skip basics, go to mechanism, trade-offs, edge cases, and where the
simple model breaks.
3. **Teach in this order, pausing where marked.** Keep each part tight.
**a. Hook (2-4 sentences).** A question or puzzle the module will let them answer,
ideally one they get wrong right now. Ask them to guess before reading on. Wait.
**b. Explanation.** Cover each objective in plain language, at most ~250 words per
objective. One diagram in a code block if structure matters. End with a one-line
summary per objective.
**c. Worked example.** A complete problem solved step by step, each step labeled
with *why* it is taken, not only what. Choose a problem that exercises the
objectives together.
**d. Faded example.** A parallel problem (same structure, new surface) where the
first steps are shown and the last 1-3 steps are blank (`Step 3: ____`). For
`advanced` learners fade everything after step 1. Ask the learner to fill the
blanks. **Wait for their answer.** Then give feedback: what is right, the exact
point where it goes wrong if it does, and one hint to fix it before showing the
correct step. If they are stuck after one hint, hand off to the same hint ladder
**learn-tutor** uses (nudge, strategy, partial step), then reveal.
**e. Retrieval questions (exactly 3).** Without the explanation in view:
1. recall: state or define a key idea in their own words,
2. apply: use it on a fresh small case,
3. connect: relate it to a prerequisite or predict what happens if something changes.
Ask all three, wait, then grade each: correct / partly / not yet, with one line of
feedback. Log each to `quiz-log.jsonl` and misses to `mistakes.md` using the
formats in **learn-quiz** (type `retrieval`).
4. **Write `learn/<slug>/lessons/m<n>-<module-slug>.md`** containing the hook, the
explanation, both examples with the correct faded steps, the three retrieval
questions with model answers, and a "Your answers" section summarizing how the
learner did. This is their notes; keep it clean and re-readable.
5. **Close.** In 3 lines: what they can now do (objectives as verbs), one thing to
watch out for, and the next step:
- any retrieval miss -> "Want to work on <miss> with the tutor?" (**learn-tutor**)
- all correct -> "Make flashcards for this module?" (**learn-flashcards**), then
"Quiz on M<n> tomorrow" (**learn-quiz**) because a delay makes retrieval stronger.
## Decision rules
- If the learner answers the hook correctly with a good reason, shorten the
explanation and spend the time on the faded example and a harder variant.
- If two of three retrieval questions are missed, do not move to the next module.
Re-teach the missed objective with a different example, then ask one new question.
- Prefer examples from the learner's own domain (learner.md preferences, the
source document, or their project) over textbook ones.
- If the learner says the explanation is over their head or too basic, ask the
**explainer** agent for the same idea at five levels, let them pick one, and
continue the lesson at that depth (and note the preference in learner.md).
- Use the syllabus source document as the authority when there is one; quote its
section names so they can go back to it.
## What NOT to do
- Do not dump the whole lesson in one message. The pauses at the hook, the faded
example and the retrieval questions are the point.
- Do not answer the faded example or retrieval questions for the learner before they
try. "I don't know" is a valid try; respond with a hint, not the answer.
- Do not praise vaguely ("Great job!"). Say what exactly was right.
- Do not cover objectives that belong to later modules; note them as "coming in M<n>".
- Do not exceed the module's time estimate by more than half; split into two sessions
instead and say where you stopped.
Wersje
W bazie nie ma innych skilli o nazwie learn-lesson.
Migawka bazy z 13.11.2026 · wersja paczki 6de5bddcae7c (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100teach
Uczy jednego tematu w głąb jako interaktywnej lekcji: intuicja, mechanizm, laboratorium z liczbami i sprawdzenie zrozumienia, także w planie nauki. - 100teach-lesson
Pisze krótką lekcję jednego pojęcia prostym angielskim, z przykładem i pytaniem do samodzielnego sprawdzenia. - 100teach
Prowadzi lekcję o wybranym temacie, zbudowaną z kanonicznych punktów kontrolnych, na przykład jak dane repozytorium obsługuje dany temat. - 90course-module-video
Zamienia plan lekcji lub notatki eksperta w wideo modułu kursu z celami nauczania, rozdziałami i podsumowaniem z pytaniem sprawdzającym. - 100answer-questions
Pełni rolę prywatnego korepetytora przy pytaniach o lekcję, ćwiczenie lub pojęcie z kursu. - 100learn-syllabus
Zadaje kilka pytań diagnostycznych i pisze plan kursu z modułami, celami, wymaganiami wstępnymi, czasem i projektem końcowym.