canvas courses export-context Specification

This is the most complex read command. It aggregates data from multiple Canvas API endpoints into a single JSON file for offline review and agent consumption.

Usage

canvas courses export-context --course COURSE_ID --out context.json
canvas courses export-context --course COURSE_ID --out context.json --include modules,assignments,files
canvas courses export-context --course COURSE_ID --out context.json --since 2026-01-01 --json

Flags

Flag	Type	Default	Description
--course	string	required	Course ID
--out	string	stdout	Output file path
--include	string	all	Comma-separated list of sections to include
--since	string	(none)	Only include items updated after this date (ISO 8601)
--download-files	string	(none)	Directory to download file attachments into
--json	bool	false	Wrap in standard JSON envelope

Sections and API calls

The command fetches sections in this order. Each section is isolated — a failure in one section does not abort others.

Section	API endpoint	Notes
course	GET /api/v1/courses/{id}?include[]=term&include[]=total_scores	Always fetched first
tabs	GET /api/v1/courses/{id}/tabs	Helps discover enabled tools
modules	GET /api/v1/courses/{id}/modules?include[]=items&include[]=content_details&per_page=100	Auto-paginate. If inline items omitted by Canvas, fallback to GET /api/v1/courses/{id}/modules/{module_id}/items
assignments	GET /api/v1/courses/{id}/assignments?per_page=100&order_by=due_at	Auto-paginate
assignment_groups	GET /api/v1/courses/{id}/assignment_groups?per_page=100
files	GET /api/v1/courses/{id}/files?per_page=100	Metadata only unless --download-files
pages	GET /api/v1/courses/{id}/pages?per_page=100 then GET /api/v1/courses/{id}/pages/{url} for each	Get body content
announcements	GET /api/v1/announcements?context_codes[]=course_{id}	Auto-paginate. (Fallback: discussion_topics?only_announcements=true if caller lacks global view permission)
discussions	GET /api/v1/courses/{id}/discussion_topics?only_announcements=false&per_page=100	Exclude announcements
submissions	GET /api/v1/courses/{id}/students/submissions?student_ids[]=all&per_page=100 or per-assignment	Only if user has permission
grades	GET /api/v1/courses/{id}/enrollments?user_id=self&include[]=total_scores	Student's own grades

Output JSON shape

{
  "course": { ... },
  "tabs": [ ... ],
  "modules": [ ... ],
  "assignments": [ ... ],
  "assignment_groups": [ ... ],
  "files": [ ... ],
  "pages": [ ... ],
  "announcements": [ ... ],
  "discussions": [ ... ],
  "submissions": [ ... ],
  "grades": [ ... ],
  "_export_meta": {
    "generated_at": "2026-06-12T19:20:00Z",
    "course_id": "123",
    "sections_requested": ["modules", "assignments", ...],
    "sections_succeeded": ["modules", "assignments", ...],
    "sections_failed": [],
    "warnings": [],
    "request_count": 15,
    "duration_ms": 3200
  }
}

Error isolation

• If a section fails with 403 (permission denied), record a warning and continue.
• If a section fails with 401, abort the entire export (auth is broken).
• If a section fails with a network error, retry per the standard retry policy, then record a warning and continue.
• The final output includes sections_failed so the consumer knows what's missing.
• Exit code: 0 if all requested sections succeeded, 8 (partial failure) if some sections failed.

--json envelope

When --json is used, the export data goes into data and the standard envelope wraps it:

{
  "ok": true,
  "data": { "course": {}, "modules": [], ... },
  "meta": { "command": "courses.export-context", ... }
}

When --json is NOT used, the raw export JSON (without envelope) is written to --out or stdout.

Concurrency and Pagination

• Concurrency: The command defaults to a concurrency of 1. It may be configurable to a very small value, but Canvas's official rate-limiting guidelines recommend single-concurrent clients for safety.
• Pagination: The implementation MUST strictly follow Canvas official pagination rules. Always inspect the Link header, and treat returned link URLs as opaque absolute URLs (do not attempt to parse or construct the next page URL manually).