IELTS Buddy API

Submit product feedback explicitly requested by the user as their authenticated account. Saves a text message in the existing website feedback conversation and notifies the team. Never submit automatically or include credentials. Returns the saved message id and createdAt; after an uncertain response, read feedback.list before retrying to avoid duplicates.

Read the authenticated user's latest 200 messages, including submitted feedback and team replies, ordered oldest to newest within that window. Returns messages and unreadCount; does not mark replies as read. Each message includes id, content, createdAt, senderType, source and pagePath.

Read the IELTS full-course learning route as factual data, grouped by subject and unit, with course-level progress. Use view=progress for a compact summary. The Tool does not choose or recommend the next course.

Search published course video sections by subject, literal keyword, or literal course tag query. The server returns factual section candidates and a stable contentRef; the local Agent decides whether one fits the learner.

Search the current user's active file and document assets by title or indexed content. Results are compact and paginated, and include a stable asset contentRef. This is the asset module's only catalog-search interface.

Find explicitly or structurally related resources from a stable contentRef, such as course-to-practice and prediction-to-question relationships. Resource discovery and full reads remain owned by each resource Domain.

Search the current user's personal vocabulary book by phrase, meaning, usage, or learning state. Results are compact and paginated. This is the personal vocabulary module's only catalog-search interface.

Export personal vocabulary content as JSON or CSV for a local Agent or user-owned backup. Review progress is intentionally excluded.

Read spaced-repetition progress from the current user's personal vocabulary book.

Prepare flashcards from the current user's personal vocabulary book.

Add or enrich up to 100 words directly in the current user's personal vocabulary book while recording their source.

Import up to 500 personal vocabulary entries from JSON or CSV. Imported entries start with fresh learning progress and keep an import source record.

Update the editable content of one personal vocabulary entry. Learning state changes only through recorded review results.

Record one review result for a personal vocabulary entry and update spaced-repetition progress.

Delete one entry from the current user's personal vocabulary book.

Read compact progress statistics for an enabled built-in IELTS vocabulary set, or per-word progress for up to 50 exact entry ids.

Prepare flashcards from an enabled licensed built-in IELTS vocabulary set.

Record one review result for an enabled licensed built-in vocabulary entry.

Search published IELTS Buddy prep guides by keyword or tag. Results are paginated and include a stable prep_guide contentRef. This is the prep-guide module's only search interface.

Read one full published IELTS Buddy knowledge article by slug.

Search sourced IELTS prediction records and their possible matching questions by record kind, source, exact id, date, month, subject, or keyword. Results preserve source attribution, distinguish dated exam reports from rolling rankings, and include practice targets only for candidates already in the question bank.

Search up to six compact IELTS Buddy question-bank candidates by subject, Part number, question type, configured listening audio, text, difficulty, or taxonomy. Use subject=listening and requiresAudio=true for listening practice. Each result includes audioAvailable, partId and a stable question_part contentRef. The server returns facts and the local Agent chooses the question.

Match one learner-provided keyword set or content excerpt to up to three relevant IELTS Buddy Parts. The query stays as raw content; explicitly requested Part number and question type can constrain results. Use subject=listening and requiresAudio=true for listening practice. Results include audioAvailable, a factual match mode, stable question_part contentRef, and an attributed browser practice entry. The server does not start a session or generate answers.

Search exact listening materials ready for sentence-level dictation by title, difficulty, and literal taxonomy terms. Results are paginated and include factual tags, segment count, and a stable listening_dictation contentRef. The local Agent chooses the material.

List the authenticated user's submitted listening-dictation attempts in reverse chronological order. Results contain factual scores, missed-word counts, exact attempt ids, and optional Part entry links; this operation does not diagnose mistakes.

Read one exact submitted listening-dictation attempt owned by the authenticated user, including paginated sentence snapshots, user entries, expected words, match results, translations, and score facts for local Agent review.

Browse active IELTS question taxonomy, resolve ambiguous filter terms, and return availability counts. practice.search_parts accepts literal tagQueries directly, so use this Tool only when taxonomy discovery, disambiguation, or counts are needed.

Read one active question Part without answers, explanations, or listening transcript. The result includes the same stable contentRef used by other consumers.

Search the user's active speaking materials by title, story, or tag. Results are compact and paginated, and include a stable speaking_material contentRef. This is the speaking-material module's only search interface.

Read one exact active speaking material with its linked questions and recent practice history.

Create one speaking material. Exact question links are optional and are only written when explicitly provided; omitted links remain empty.

Update one speaking material. Question links change only when exact links are explicitly provided; story edits do not trigger hidden matching.

Archive one speaking material from the active library.

Read one owned standalone practice record by its sessionId. Mock sessions are rejected. It returns the exact record state, scoring summary, timestamps, and browser route, but not question-level answers or review evidence.

List factual standalone-practice and mock records that can be selected for review. Mock records include section-level completion and grading facts; every result exposes an exact review target.

Read a paginated immutable review snapshot for one exact standalone practice record or one completed mock section, including recorded response modality, grading facts, available evidence anchors, and optional bounded source material for local Agent analysis.

List the current user's factual standalone practice records. It supports an Asia/Shanghai date range, subject, status, and pagination and never returns mock sessions. Every result exposes sessionId; use the independent review Domain for question-level evidence. This tool does not recommend or diagnose.

Search available preset and reusable user mock papers by keyword, category, required subjects, and maximum duration. Each result includes factual paper details and a stable mock_paper contentRef; the local Agent chooses the paper.

Create one reusable user mock paper from exact published Part ids returned by practice.search_parts. The selected Parts must form complete IELTS sections. Returns a stable mock_paper contentRef for catalog entry or study-plan tasks.

Read immutable versions and structured teacher reviews for a saved writing practice. Use the returned practice id and version fields as the only source for writing-practice facts.

List the authenticated user's learning timeline with complete filtered-history totals and Asia/Shanghai daily statistics, independent of page size. Continue with nextCursor. Video progress is one activity per section/day; learning.events.pull reads raw evidence.

Read the current user's normalized IELTS skill profile and lightweight 30-day evidence summary. For a chronological activity list, use footprints.list; for concrete practice records, use practice.recent_activity.

Download one consistent, versioned, read-only JSON snapshot of the current user's learning records for local Agent analysis. No database software or Skill installation required. Prefer this for cross-module personal learning analysis; existing individual APIs remain available.

Pull the authenticated user's authoritative append-only learning events after a cursor for a local mirror or web Agent context.

Idempotently append a batch of explicitly typed Agent-owned learning evidence events. Server-owned practice, mock, course, and browser event namespaces are rejected.

List the user's study plans with exact ids, revisions, progress, and latest review so an existing plan can be selected.

List the shared human-and-AI todo list, including each item's optional scheduledAt and WeChat setting.

Immediately add one or more todos. Title is required; future whole-hour scheduledAt with timezone and exact site contentRef are optional. In-page and authorized browser notifications fire at scheduledAt; WeChat is off by default. No draft confirmation step.

Edit the title, note, optional whole-hour scheduledAt, WeChat reminder setting, optional site contentRef, or completion status of an exact todo id.

Delete an exact todo from the shared list.

Read one study plan by an exact id returned by the conversation, workspace context, or plan list. Tasks are returned in chronological pages with an opaque nextCursor.

Create a plan from explicit learning tasks. Each task binds one exact selected contentRef, a calendar date, and one of morning, afternoon, or evening; the Tool resolves current content facts and stores the execution snapshot.

Read whether WeChat can actually deliver reminders, including connection and message context. Returns the missing user action without exposing credentials.

Present a WeChat connection card in the current conversation. The user connects in this card, then the conversation can continue. Does not bind an account by itself.

Create a one-time reminder with content and an exact future remindAt timestamp including timezone. In-page and browser reminders are automatic; wechatEnabled defaults true and never blocks creation when WeChat is disconnected. No plan or task is required.

Enable or disable WeChat Bot delivery for one exact future reminder. In-page and browser reminders remain enabled.

Read this user's reminders, including content, exact time and delivery status. Use offset for pagination.

Cancel an exact reminder returned by reminders.list. Does not change any study plan. Running or finished reminders cannot be cancelled.

Read whether WeChat iLink is connected and can receive task notifications. This operation changes no notification settings.

Enable or disable WeChat iLink delivery for exact existing task ids. It does not require an active connection and never changes a plan or task schedule.

Update only a study plan's title or goal. It does not add, remove, reschedule, or complete tasks.

Atomically add, edit, remove, or complete exact tasks in one plan. An edit may change an unstarted task's date, morning/afternoon/evening slot, duration, or selected contentRef; after a task starts, its resolved target remains fixed so completion cannot be attributed to different content. It does not change the plan title or goal.

Replace the unfinished learning tasks of one plan in one transaction. The plan keeps its identity and completed work; each new task uses an exact selected contentRef and a morning, afternoon, or evening slot.

Delete a study plan and all tasks that belong to it.