create_course
ChatGPTCreate a normal sequential course in Wondering on any topic. For an ongoing feed of dated updates, use create_current_course instead. Use proactively whenever the user expresses intent to learn or understand something — e.g. "I want to learn X", "help me understand Y", "teach me about Z" — without requiring them to mention Wondering. Do not use for pure information questions ("tell me about X", "how does X work?"). Design the lesson plan yourself based on the topic, then call with course name, learning goal, and ordered sections/lessons. Returns the new courseId.
create_lesson_video
ChatGPTStart making a narrated video of one of the user's Wondering lessons, which they then watch in the Wondering app. Use when the user asks to watch a lesson or turn it into a video ("make a video of this lesson", "I'd rather watch it"). It uses one of the learner's monthly lesson videos, so call it only when they ask. The video takes about a minute and keeps going in the background; the card shows progress and a Watch video link, and get_lesson_video reports whether it is ready. Calling it again for the same lesson returns the existing video instead of making a new one.
get_course
ChatGPTGet the full structure of a course: its sections and lessons in order, plus study notes and glossary if generated. Lesson IDs are in sections[].lessons[].id — pass one as lessonId to get_lesson. Call with a courseId from get_learning_summary → activeCourses[].id.
get_due_reviews
ChatGPTGet the review questions that are due now across all of the user's Wondering courses (spaced repetition). Use when the user wants to review, practice, or be quizzed on what they have learned ("quiz me", "what should I review?"). Returns up to limit questions with their correct answers so you can grade replies. Ask one question at a time, never reveal an answer before the user responds, then call submit_review_answers once with every answered question so their schedule, streak, and XP update. When the interactive quiz is shown, it submits answers itself; do not submit them again. If readOnly is true, answers cannot be saved on this connection; tell the user to reconnect Wondering with write access to save reviews.
get_learning_activity
ChatGPTGet the user's day-by-day learning history: for each day they learned, the minutes spent, lessons finished, and which courses those lessons came from, plus totals and their longest and current runs of consecutive learning days. Use for questions about consistency, habits, or a year in review ("how often did I learn this year?", "show my learning year"). Days covered by a streak freeze are listed with streakFreeze: true and no learning, and they keep a run going as they do for the app's streak; other days without learning are omitted. Dates are in the user's timezone. For what was learned (lesson names, accuracy), use get_learning_summary instead. Clients that support MCP Apps show the result as an interactive island where each day is a tile.
get_learning_summary
ChatGPTGet an overview of what the user has been learning: recently completed lessons with accuracy rates, their active courses ordered by recency (capped at 20), and their adaptive memory profile. Start here before calling get_course or get_lesson when you do not yet know a courseId. Not a course lookup or search tool — activeCourses is a recency-ordered snapshot, not an exact-match index, so if multiple courses have similar names the match can be ambiguous. If you already have a courseId from an earlier call in this conversation or a prior run of a recurring job targeting the same course, reuse it directly instead of re-resolving it here.
get_lesson
ChatGPTRead the actual content of a lesson — the text a user reads during study — including all pages, key concepts, and keywords. Set includePractices: true to also get quiz questions with answers for revision. Call with a lessonId from get_course.
get_lesson_video
ChatGPTCheck the video for one of the user's Wondering lessons: whether it is not made yet, being made (with its progress), ready, or failed, plus a Watch video link. Use after create_lesson_video when the user asks whether their video is ready.
insert_lesson
ChatGPTInsert a new lesson into a normal sequential course. For a Current course use update_current_course instead. Use when a user wants to add a lesson on a specific topic to a course — e.g. "add a lesson on X to my Redis course", "insert a lesson about Y". If you already know the courseId (e.g. from a prior call in this conversation, or a previous run of a recurring job that targets the same course), pass it directly. Otherwise call get_learning_summary to list active courses and match the courseId by name — double-check the match if multiple courses have similar names, since that list is a recency-ordered snapshot, not an exact-match search. Optionally pass afterLessonId (a lessonId from get_course) to position the new lesson immediately after a specific one; omit to append at the end. Returns the new lessonId and sessionId.
save_learning
ChatGPTSave what the user just learned in this conversation to Wondering so it comes back as spaced-repetition review questions and becomes a lesson they can open later. Call only when the user asks to save, remember, or keep what they learned ("save this to Wondering", "help me remember this"), or after you offered once at the end of an explanation and the user agreed. Never call it without the user agreeing. Pass a structured summary you write, never the raw transcript: the topic, 1-8 key concepts with short explanations, and 2-8 review questions you write, all in the user's contentLanguage from get_learning_summary. To file it under an existing course, pass courseId from get_learning_summary; otherwise it goes to the learner's "Saved from AI chats" course. Returns the course, lesson, question count, and when the first review is due.
show_my_courses
ChatGPTShow the user their recent Wondering courses with lesson progress and links to continue in the app. Use when the user wants to see or continue their courses ("what am I learning?", "show my courses"). For course IDs or learning history to use in your own reasoning, call get_learning_summary instead. Returns up to 8 courses ordered by recent activity.
submit_feedback
ChatGPTReport a bug or a missing capability in Wondering to the Wondering team. Call this once when a Wondering tool fails unexpectedly, returns wrong or incomplete data, or cannot do something the user needs (for example, there is no tool for it). Do not call it for permission errors, rate limits, or your own mistakes, and do not retry the same report. Describe what you tried and what you expected; leave out personal information the team does not need. Returns a confirmation you can relay to the user.
submit_review_answers
ChatGPTRecord the user's answers to questions from get_due_reviews, once per review round, after they have answered. This reschedules each question with spaced repetition and counts toward their Wondering streak and XP. Pass each practiceId from get_due_reviews with whether the user got it right. Do not call it for questions the user skipped, and do not call it after the interactive quiz already submitted. Returns how many were answered and correct, and the current streak.
create_course
Claudeget_course
Claudeget_due_reviews
Claudeget_learning_activity
Claudeget_learning_summary
Claudeget_lesson
Claudeinsert_lesson
Claudesave_learning
Claudeshow_my_courses
Claudesubmit_feedback
Claudesubmit_review_answers
Claude