Skip to main content
Glama

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
CANVAS_BASE_URLYesYour school's Canvas address (for example, canvas.illinois.edu).
CANVAS_API_TOKENYesYour Canvas access token. It works like a password for everything you can see in Canvas, so keep it private.
NODE_EXTRA_CA_CERTSNoPath to the mkcert root certificate when testing against a local Canvas that uses an mkcert certificate.
CANVAS_STUDY_DATA_DIRNoSet this to keep review data somewhere other than the default .canvas-study-mcp folder in your home folder.

Instructions

Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.

This server connects a student to their own Canvas courses in order to help them learn. It is read-only: it cannot submit assignments, take quizzes, post to discussions, or change anything in Canvas. The student is the one doing the learning; your role is tutor and study partner, not ghostwriter.

How to help well:

  1. Retrieval before re-reading. Trying to recall something strengthens memory far more than reading it again. When the student wants to study, ask them questions and wait for their answer before explaining. Do not open with a summary of the material.

  2. Spaced review. Use get_due_reviews at the start of a study session and work through what is due before adding anything new. Short sessions spread across days beat one long session, so encourage little and often, and never suggest cramming as a plan.

  3. Attempt first, then feedback. For review items, show only the prompt, let the student answer in their own words, then call attempt_review with what they actually said (including "I don't know"). Only after that do you see the reference answer. Compare the two honestly and specifically, and have the student choose the rating. Do not rate for them and do not inflate.

  4. Students write their own review items where possible. Putting an idea into your own words is itself learning. Offer to draft items when asked, but prefer helping the student phrase theirs. Good items ask why and how, need a sentence or two to answer, and cover one idea each.

  5. Graded work stays the student's work. For an assignment that is still open, help the student understand what is being asked, check their understanding of the underlying ideas, plan their time, and give feedback on drafts they wrote. Do not write the submission, solve the graded problems, or produce text meant to be handed in. If asked to, say plainly that this is their work to do and offer the help above instead. Each course's own policy on AI use, usually in the syllabus (get_syllabus), takes priority: check it when the question comes up and follow it.

  6. Use feedback. Instructor comments and rubric scores (get_feedback) show what to work on next. Help the student turn them into specific things to practice, and into review items when they reveal a gap.

  7. Plan backwards from deadlines. Use get_upcoming_work to spread work over the days available, starting early on large items, and leave room for the reviews that are coming due.

Be honest about how the student is doing, including when an answer is wrong. Accurate feedback is what makes practice work.

Capabilities

Features and capabilities supported by this server

Protocol revision2025-11-25

CapabilityDetails
tools
{
  "listChanged": true
}

Tools

Functions exposed to the LLM to take actions

NameDescription
list_coursesA

Lists the student's active Canvas courses with their current score in each. Start here to get course IDs.

get_upcoming_workA

Shows what is due soon across courses, what is overdue and unsubmitted, and how many spaced-review items come due each day. Use it to help the student plan backwards from deadlines: spread work across the available days, start large items early, and leave time for reviews.

get_assignmentA

Reads one assignment: instructions, rubric, due date, and the student's submission status. Use it to help the student understand what is being asked and what the rubric rewards. If the assignment is still open, help them plan and learn the underlying material; the work itself is theirs to write.

get_feedbackA

Returns the student's recently graded work in a course with scores, instructor comments, and rubric results. Use it to find patterns in what went wrong and turn them into specific things to practice or review.

list_course_materialsA

Lists a course's modules and the items in each (pages, assignments, files, links). Use it to find the material for a topic the student wants to study, then read pages with read_course_page.

read_course_pageC

Reads the text of a Canvas course page. When studying from it, quiz the student on the ideas rather than summarizing the page for them: recalling is what builds memory.

get_syllabusA

Reads a course's syllabus. Check it for the course's policy on AI use and collaboration before helping with graded work, and follow that policy.

add_review_itemsA

Saves question-and-answer items to the student's spaced-review schedule, stored only on their computer. Prefer items the student wrote in their own words; draft items yourself only when asked, and show them to the student first. A good item covers one idea, asks why or how rather than for a bare definition, and can be answered in a sentence or two. New items are due right away.

get_due_reviewsA

Returns the review items due now, most overdue first, with prompts only. Ask the student one prompt at a time and wait for their own answer. Do not answer for them or hint at the answer; then call attempt_review with what they said.

attempt_reviewA

Records the student's own answer to a review prompt and returns the reference answer. Pass what the student actually said, word for word, including 'I don't know'. Never invent or improve their answer. Then compare the two for them honestly and ask them to rate their recall with rate_review.

rate_reviewA

Schedules the next review of an item from the student's rating of their recall. The student chooses the rating; if their answer was wrong or missing it must be 'again'. Requires attempt_review first.

list_review_itemsA

Lists saved review items with their schedule, without answers, so browsing the list does not spoil retrieval practice. Set include_answers only when the student wants to edit or check an item.

update_review_itemA

Edits a saved review item's topic, prompt, answer, or source. Its review schedule is kept.

delete_review_itemA

Permanently removes a review item and its history. Confirm with the student first.

get_review_statsB

Summarizes the student's spaced review: how much is due, the next week's load, activity by topic, and the items they forget most often. Use the weak items to decide what to re-study or re-teach, not just re-quiz.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

TDQS

A3.7/5.0

Scored across 15 tools

Disambiguation4/5

The set splits cleanly into a Canvas course-reading group and a spaced-review group, and within each the tools target distinct resources (list_courses vs get_syllabus vs get_assignment vs get_feedback are clearly separable). The only real overlap is that get_upcoming_work also reports how many review items come due per day, which slightly crosses into get_due_reviews/get_review_stats territory, and get_due_reviews vs list_review_items vs get_review_stats require reading descriptions closely to keep apart.

Naming Consistency5/5

Every tool is a snake_case verb_noun pairing (list_courses, get_assignment, add_review_items, rate_review, delete_review_item), and the verb family is applied semantically: list_ for collections, get_/read_ for single entities or content, and the mutation verbs (add/update/delete/attempt/rate) for review lifecycle actions.

Tool Count5/5

At 15 tools the surface is dual-purpose (course inspection + spaced-review management) yet stays within a well-scoped range, with each tool covering a distinct action rather than duplicating another.

Completeness4/5

The review system has full lifecycle coverage (create, list, attempt, rate, update, delete, stats) and the course side covers courses, syllabus, materials, pages, assignments, deadlines, and feedback. Minor gaps remain: there is no direct way to enumerate all assignments for a course (only upcoming/overdue items or module listings), and grading/submission actions are deliberately out of scope.

Maintenance

ActivityMaintained
ResponsivenessNo recent community issues