Preview or incrementally sync course files into semester/course/original folders. Defaults to dry_run=true. Uses a manifest and checksums, preserves local edits, never deletes files. Inspect errors and truncated before claiming a complete sync.

Aggregate gradebook, calendar and assignment pages with source links, exact-match merging, explicit submission status and conflicting deadlines. Includes separate unverified deadline excerpts from announcements and assignment documents; never treat excerpts as confirmed deadlines. Check errors/truncated/documentScan for coverage.

List courses the current user is enrolled in on NTULearn. By default returns only active/available courses. Set include_disabled=true to also include unavailable ones. Supports pagination via limit/offset.

Walk a course's content tree. Omit parent_id to get the top-level items (folders, documents, links, assignments); pass parent_id of a folder/lesson where hasChildren=true to drill into its children. Supports pagination.

Recursively search a course's entire content tree for items matching a query. Matches on title or description (case-insensitive substring). Returns matched items with their full breadcrumb path.

Download every file attached to a Blackboard content item to local disk. Handles both resource/x-bb-file (attachment API) and resource/x-bb-document (HTML body with bbcswebdav links) handler types. Pass destination_dir to target a specific folder — useful for organising a semester (e.g. destination_dir='~/NTU/y3s1/sc2002/week 8/'). Returns saved files with their resolved local paths and sizes. Use ntulearn_read_file_content if you want to inspect the content inline rather than saving to disk.

Read the content of files attached to a Blackboard content item, returned inline (no local-filesystem hop). Use this to ask questions about lecture material — ntulearn_download_file is for users who actually want the bytes on disk. PDFs default to text mode (via pypdf — cheap and almost always what you want for written content). Pass mode='vision' for diagram-, equation-, or screenshot-heavy pages; pair with pages='5' or pages='1-3' to keep the payload under MCP's 1 MB cap (~3K vision tokens per page). For multi-page diagram-heavy decks, prefer ntulearn_download_file plus drag-and-drop into claude.ai. Also supports Microsoft Office formats (.docx, .pptx with speaker notes, .xlsx with all sheets) and text-like files (txt, md, csv, json, xml, html with tags stripped, code files). Other binaries (images, video, audio, archives, legacy .doc/.ppt/.xls) are listed under skipped — fall back to ntulearn_download_file for those. Per-file cap 25 MB, batch cap 40 MB, vision cap 50 rendered pages; oversized files are skipped.

Get upcoming calendar items and assignment due dates across your enrolled courses. Wraps Blackboard's calendar API — assignment due dates surface as items with type='GradebookColumn'. By default returns the next 2 weeks across every available course (server fans out per-course in parallel). Pass course_ids to scope to specific courses, since/until (ISO-8601) to override the window, or type to filter (e.g. type='GradebookColumn' for due dates only).

Get announcements across your enrolled courses, newest first. By default fans out across every available course; pass course_ids=['_123_1'] to scope. Use since (ISO-8601) to filter to recent announcements only (e.g. "this week"). Each item includes the courseId it was posted to so cross-course views stay attributable. Supports pagination.

Get gradebook columns across your enrolled courses, including your scores where available. By default fans out across every available course; pass course_ids=['_123_1'] to scope. Each column carries the courseId it belongs to so cross-course views stay attributable. Supports pagination.