agentleFS
liknenSliknenS/language-learning-skills/skills/vocab-coach/SKILL.md

vocab-coach

Skill0 starsChanged 27 days agoMITView on GitHub

Vocabulary coach for a foreign language - picks the words and phrases worth learning from any text, chat, email or conversation, saves them and reviews them with spaced repetition. Use this skill whenever a language learner asks which words to learn from a text, wants words explained and kept, asks for flashcards, a word list or a vocabulary review, or pastes text in the language they study - even if they do not say "flashcards". Also for "добавь слова в словарь", "які слова вивчити", "zrób fiszki".

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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. Distinctive:

    Review: Ask from memory first (meaning → term), then by recognition.

    line 22
  2. Very specific:

    Start with today() and profile() if you have not called them this session: they give the course language, level, goal and ui_lang (the …

    line 38
  3. Very specific:

    Use topics() to see the course decks by level and vocab_list(tag, skill_below) to find words the learner does not recall confidently yet.

    line 43
  4. Very specific:

    If you can write files (Claude Code, an agent with a workspace), keep vocab.csv with columns term,meaning,example,added,next_review,grade …

    line 52

≈ “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

Vocabulary coach

Pick the words and phrases worth learning from whatever the learner brings, store them, and bring them back on a schedule until they stick.

Steps

  1. Know the learner. Course language, level (A1-C2), goal (work, exam, relocation, travel) and the language to explain in. Ask once if you do not know them.
  2. Pick, don't dump. From the source take 5-15 items that match the level and the goal. Prefer chunks the learner will say (take a day off, zależy mi na tym, man reikia) over rare single words. Skip what is clearly known.
  3. One entry, one thing. A word in its dictionary form, with the form from the text kept as the example. Or one whole ready phrase. No ..., no /, no templates with gaps; split lists (days, numbers) into separate entries.
  4. Show before saving. A short table: term, meaning in the explanation language, example. Let the learner drop or add items, then save.
  5. Review. Ask from memory first (meaning → term), then by recognition. Grade each answer 1 didn't recall, 2 hard, 3 recalled, 4 easy, and space the next review: again in the session for 1-2, a few days for 3, a week or more for 4.

Example entries (Polish course, explained in Ukrainian):

termmeaningexample
umówić sięдомовитися (про зустріч)Umówmy się na wtorek.
na razieпоки щоNa razie mieszkam u znajomych.

If Talkfirst MCP is connected

Talkfirst tools are available when you see tools such as today, vocab_add, srs_due (your host may add a prefix to the names).

  • Start with today() and profile() if you have not called them this session: they give the course language, level, goal and ui_lang (the language to explain in). Read docs("vocab") before adding words: it has the entry rules the server expects.
  • Save with vocab_add(items) (term, meaning, example, level, kind: chunk up to ~5 words or formulation for a whole sentence). Duplicates are skipped by the server.
  • Use topics() to see the course decks by level and vocab_list(tag, skill_below) to find words the learner does not recall confidently yet.
  • Review with the cards: srs_due() for what is due, srs_review(card_id, grade) with the 1-4 grade. For a review in a window, quiz_start(source="due"); the quiz window sends its result to the chat itself, do not prompt answers.
  • Learner says "postpone" → defer_card; "never learn this" → skip_card; "I already know it" → mark_known.
  • The words come back in the Talkfirst app (web and Telegram) and in voice conversations on the same account.

Without it

  • If you can write files (Claude Code, an agent with a workspace), keep vocab.csv with columns term,meaning,example,added,next_review,grade and update it after every review. Otherwise end the session with the table as tab-separated text the learner can import into Anki or any flashcard app.
  • At the start of the next session ask the learner to paste the list back, then review what is due.

Persistent vocabulary with spaced repetition, quizzes and the same words in voice practice: connect Talkfirst (free plan, MCP connector for Claude and ChatGPT).

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.