teach-back
/teach-back
Sprawdza rozumienie, prosząc użytkownika o wyjaśnienie tematu własnymi słowami, i zwraca luki w formie pytań, nic nie pisząc.
- Autor
- erayozenn ↗
- Repozytorium
- BigHead-2ndBrain ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 10.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 1
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 10.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Sprawdza rozumienie, prosząc użytkownika o wyjaśnienie tematu własnymi słowami, i zwraca luki w formie pytań, nic nie pisząc.
Argumenty
/teach-back <topic or notes/ file>
Oryginalny opis (EN)
Check the user's understanding by having THEM explain a topic in their own words (Feynman technique) — compare their explanation with the vault's wiki and return gaps, errors and vague spots as questions, without writing anything. Use when the user wants to test themselves, says they will explain or teach a topic back, or shares their own explanation or note for feedback. Not for requests where the user wants Claude to explain something.
Jak działa
- 1. Get the explanation
- 2. Find the reference
- 3. Compare
- 4. Feedback
- 5. Close
Podgląd SKILL.md
---
name: teach-back
description: Check the user's understanding by having THEM explain a topic in their own words (Feynman technique) — compare their explanation with the vault's wiki and return gaps, errors and vague spots as questions, without writing anything. Use when the user wants to test themselves, says they will explain or teach a topic back, or shares their own explanation or note for feedback. Not for requests where the user wants Claude to explain something.
argument-hint: "<topic or notes/ file>"
allowed-tools:
- Read
- Grep
- Glob
- Bash(wc -c *)
---
# Teach-back
Arguments: `$ARGUMENTS` — a topic, or the path of the user's own note under `notes/`.
The vault's `CLAUDE.md` is the authority for rules (citation, markers, autonomy). Talk in the
vault's chat language. This skill **writes nothing**: the explanation is the user's work, and
`notes/` / `author: me` files are never edited (the guard hook also blocks it).
**Preconditions.** The current directory is a vault: `CLAUDE.md` and `index.md` exist. If not,
stop and say so.
## 1. Get the explanation
- A `notes/` path: Read it. That is the explanation.
- A topic: ask the user to explain it in the chat, in their own words, as if to a classmate —
without looking at the wiki first. Suggest 5–10 sentences. Wait.
- Do not explain the topic yourself before they have tried.
## 2. Find the reference
Read `index.md`, then find the wiki pages for the topic as the query skill does (terms in both
languages, `aliases:`, then body). Budget: 48 KiB over the pages read (`wc -c` first). If the
wiki has no page for the topic, say so: feedback is then "(not in the vault)" general knowledge
only, and suggest a source to ingest.
## 3. Compare
Check the explanation against the pages. Collect at most 5 findings, most important first:
| Kind | Meaning |
|---|---|
| Error | contradicts a wiki claim (cite the page) |
| Gap | a key idea from the pages is missing |
| Vague | a term used without saying what it means, a step skipped, "it just works" |
| Beyond the vault | a claim the wiki does not cover — not wrong, just unchecked |
A claim the wiki marks (disputed) or (provisional) is not an error if the user states one side;
mention the marker instead.
## 4. Feedback
1. One or two sentences on what the explanation gets right (specific, not generic praise).
2. The findings, each phrased as a **question** that leads them to fix it themselves
("What happens to training when the learning rate is too large?"), with the page to
check (`[[page]]`). Do not write the corrected explanation.
3. Ask them to answer or revise. On the second round, confirm what is now right and give the
direct answer for anything still missing, cited.
## 5. Close
- Offer: if they want to keep the improved version, they write or update their own note in
`notes/` (Claude does not).
- A gap in the **wiki** itself (topic missing or thin) → suggest an ingest; offer to add it to
`Threads.md` (ask first — it is not free to write).
- No log line; nothing is written.
Wersje
| Repozytorium | Wynik | Licencja |
|---|---|---|
| Ta strona · erayozenn/BigHead-2ndBrain | 90 | MIT |
| dotslashsimran/simcraft | 100 | MIT |
Migawka bazy z 13.11.2026 · wersja paczki f9840a2b5995 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 1 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100feynman-technique
Prowadzi użytkownika przez technikę Feynmana: tłumaczenie prostymi słowami, szukanie luk i uproszczenie wyjaśnienia. - 95feynman-validator
Sprawdza, czy uczący się potrafi wyjaśnić pojęcie własnymi słowami i wskazuje luki ukryte pod żargonem. - 100feynman-notes
Tłumaczy pojęcie prostym językiem, wykrywa luki w rozumieniu, wraca do źródła i proponuje uproszczoną wersję. - 96teach
Wyjaśnia prosto fragment pracy: sprawdza, jak działa i dlaczego jest taki, a potem składa z tego jedno wyjaśnienie. - 96teach
Pomaga naprawdę zrozumieć temat: opisuje mechanizm i uzasadnienie, a potem łączy je w jedno czytelne wyjaśnienie. - 96teach
Zbiera w jedno czytelne wyjaśnienie wiedzę o tym, jak coś działa i dlaczego, dla osoby, która chce to zrozumieć.