Canvas Info

Canvas Course Information Research

Overview

The canvaslms CLI provides read-only access to Canvas LMS course data: assignments, pages, modules, discussions, announcements, grades, quizzes, and calendar events. This skill teaches a research workflow for answering user questions about their courses.

For detailed options on any subcommand, run canvaslms <subcommand> -h.

Research Workflow

Follow these steps when answering a Canvas-related question:

Step 1: Identify the Course

canvaslms courses              # List all courses
canvaslms courses "regex"      # Filter by name/code (case-insensitive regex)

Note the course ID from the output. Use it with -c <id> in subsequent commands.

Step 2: Explore Course Structure

Choose the right entry point based on the question:

Step 3: Drill into Content

Use view subcommands to read specific items:

canvaslms modules view -c <id> -m "module regex"
canvaslms pages view -c <id> -p "page regex"
canvaslms assignments view -c <id> -a "assignment regex"
canvaslms discussions view -c <id> -d "discussion regex"

The view commands output full content (markdown for pages, HTML for assignments). The list commands output tab-delimited CSV summaries.

Step 4: Refine with Filters

Most commands accept regex filters and additional flags. Run canvaslms <subcommand> -h to discover available filters for each command.

Common refinement patterns:

• Combine course + item filters: -c "DD2520" -a "lab"
• Filter by module: -c <id> -m "week 3"
• Filter submissions by user: canvaslms submissions -c <id> -u "username"

Step 5: Synthesize and Answer

After gathering data:

• Quote relevant content directly when answering
• Summarize lists into readable format (tables, bullets)
• If data seems stale, suggest re-running with --no-cache
• Provide the course/assignment/page names so the user can navigate in Canvas

Quick Reference

Regex Filtering Patterns

• Course matching is case-insensitive: "dd2520" matches "DD2520"
• Use anchors for precision: "^DD2520$" for exact match
• Partial match works: "crypt" matches "Applied Cryptography"
• Combine flags for specificity: -c "DD2520" -m "week 3" -a "lab"

Output and Caching

• list commands produce tab-delimited CSV (pipe through column -t -s$'\t' for readability)
• view commands produce full content (markdown, HTML, or YAML frontmatter + content)
• Results are cached; use --no-cache to force a fresh fetch
• Pages view output includes YAML frontmatter with metadata

Common Pitfalls

Always identify the course first:

BAD:  canvaslms assignments list -a "lab 1"     # No course specified
GOOD: canvaslms assignments list -c 12345 -a "lab 1"

Use list before view to discover item names:

BAD:  canvaslms pages view -c 12345 -p "syllabus"   # Guessing the page name
GOOD: canvaslms pages list -c 12345                   # Discover available pages
      canvaslms pages view -c 12345 -p "Course Syllabus"  # Use actual name

Use the right content type flag for discussions:

BAD:  canvaslms discussions list -c 12345              # Missing type
GOOD: canvaslms discussions list -c 12345 -t announcement
GOOD: canvaslms discussions list -c 12345 -t discussion

Scope

This skill is read-only — it retrieves and presents course information. It does not create, modify, or delete any Canvas content. To create quiz content, use the canvas-quiz skill instead.

Testing Write Commands

When a task legitimately requires testing canvaslms write commands (outside this skill's scope), use the user's sandbox course — see the kth-data skill (§1 "Sandbox dbosk", select with -c "^Sandbox dbosk$") for the rules: anchored regex, recognizable throwaway names, verify with --no-cache, restore afterwards, never write-test fan-out paths. The same skill is where to look up KTH rooms, period dates and the user's schedule.