Use when a learner needs an agent-maintained vocabulary notebook built from words encountered in real content.

Opens a fresh example using the template’s built-in demo data. No account required.

vocab-notebookrect.sh

What it changes

A shared surface for the whole job.

Give the work a durable interface instead of leaving it in chat.

Keep people and agents aligned around the same current state.

Share the result as one hosted, interactive page.

Best for

Reusable agent workflowsCollaborative reviewClient delivery

Key capabilities

The tutor agent banks a new word it mined from something the learner consumed. Give term and meaning; add pronunciation, pos (part of speech), an example (the actual sentence the word appeared in - the context is how vocabulary sticks), an optional note, tags, and source { title, type (drama|article|book|conversation|podcast|movie|pdf|other), detail } so the learner remembers where they met the word. New words start at Leitner box 1 (due for review). Matched by id; re-adding an existing id updates its content but preserves the learner's review progress unless those fields are passed.
Star or unstar a word — the learner flags words they want to focus on. Pass starred:true to star, false to unstar. Starred words can be filtered to on their own in the notebook. Leaves review progress untouched.
Record a finished review session for the history strip: how many words were reviewed and how many were remembered. The view passes the timestamp (at, ms since epoch).

Try asking your agent