CanvasLMS - API

CanvasLMS - API is a local MCP server that puts your Canvas LMS account in front of an AI assistant. It runs on your machine, talks to Canvas with your own personal access token, and exposes about 100 tools covering courses, assignments, grading, discussions, modules, pages, files, messaging, peer reviews, and accessibility. Every tool returns Markdown. Every tool that changes Canvas previews its change first and only applies it once you confirm.

Quick start

1. Get a Canvas personal access token: in Canvas go to Account > Settings > Approved Integrations > New Access Token.

2. Clone the repository and install it.

   git clone https://github.com/harryx2011-boop/CanvasLMS-API.git
   cd CanvasLMS-API
   uv venv .venv
   uv pip install -e .
   

   Without uv, use a standard virtual environment instead:

   python -m venv .venv
   .venv\Scripts\activate   # Windows
   source .venv/bin/activate # macOS/Linux
   pip install -e .
   

3. Copy the example environment file and fill in your Canvas host and token.

   cp .env.example .env
   

4. Verify the connection.

   canvaslms-api --test
   

   This prints your name, user id, and active courses. If it fails, check CANVAS_URL and CANVAS_TOKEN in .env.

Connect to Claude Code

Windows:

claude mcp add --scope user canvaslms-api -- "<absolute path>\.venv\Scripts\canvaslms-api.exe"

macOS/Linux:

claude mcp add --scope user canvaslms-api -- "<absolute path>/.venv/bin/canvaslms-api"

The server reads .env from the repository folder at startup, so no environment variables need to go in the client configuration. Use the absolute path to the executable inside .venv.

Connect to Claude Desktop

Add an entry to your Claude Desktop configuration file:

{
  "mcpServers": {
    "canvaslms-api": {
      "command": "<absolute path>\\.venv\\Scripts\\canvaslms-api.exe"
    }
  }
}

On macOS/Linux, use <absolute path>/.venv/bin/canvaslms-api for command.

Config file locations:

• Windows: %APPDATA%\Claude\claude_desktop_config.json
• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Restart Claude Desktop after editing the file.

Other clients

Any MCP client that supports stdio servers can launch the executable directly, the same way as Claude Code and Claude Desktop above.

For clients that only speak HTTP, run the server with the HTTP transport:

canvaslms-api --transport http --host 127.0.0.1 --port 7100

Point the client at http://127.0.0.1:7100. The server still reads Canvas credentials from .env in the repository folder.

How it behaves

• Course identifiers. Any course argument accepts a numeric Canvas id, a course code (ENG101), part of a course name, or sis_course_id:X. Use list_courses first if you are not sure which one you want.
• The confirm gate. Every tool that changes Canvas takes a confirm: bool = false argument. Called without it, the tool returns a preview of what would change and makes no request that alters Canvas. Call it again with confirm=true after reviewing the preview.
• Permissions. Every tool runs as the token owner. A student token cannot see other students' private data or use educator tools such as grading or bulk messaging; those calls return the Canvas 403 with a hint about what permission is missing.
• Output. Every tool returns Markdown, formatted for direct display in a chat transcript.
• Anonymization. When CANVAS_ANONYMIZE_STUDENTS=true, student names in tool output are replaced with a stable per-course pseudonym instead of their real name.

Tool reference

Write tools (marked confirm) return a preview until called again with confirm=true.

Courses & identity

Tool	Purpose	confirm
get_profile	Fetch the token owner's Canvas profile
get_enrollments	List the token owner's enrollments across courses
list_courses	List and search courses the token owner can see
get_course	Fetch details for one course
get_syllabus	Fetch a course's syllabus body
get_course_overview	Summarize a course: syllabus, modules, upcoming work
get_course_structure	Fetch a course's full modules/pages/assignments outline
get_cache_status	Show what course-resolution data is cached
clear_cache	Clear the course-resolution cache

Student

Tool	Purpose	confirm
get_upcoming_assignments	List assignments due soon across courses
get_todo	Fetch the Canvas to-do list
get_submission_status	Show submitted/missing/graded status for a course
get_grades	Fetch current grades for one or all courses
get_pending_peer_reviews	List peer reviews still owed by the token owner
get_submission	Fetch one submission's detail and feedback
submit_assignment	Submit an assignment	confirm
comment_on_submission	Add a comment to a submission	confirm
mark_module_item_done	Mark a module item complete	confirm

Assignments & grading

Tool	Purpose	confirm
list_assignments	List a course's assignments
get_assignment	Fetch one assignment's detail
list_submissions	List submissions for an assignment
get_assignment_analytics	Summarize submission and grade distribution for an assignment
get_student_analytics	Summarize one student's activity in a course
create_assignment	Create an assignment	confirm
update_assignment	Update an assignment	confirm
bulk_grade_submissions	Grade multiple submissions in one call	confirm
grade_with_rubric	Grade a submission against a rubric	confirm

Rubrics

Tool	Purpose	confirm
list_rubrics	List a course's rubrics
get_rubric	Fetch one rubric's criteria
get_rubric_assessment	Fetch a submission's rubric assessment
create_rubric	Create a rubric	confirm
create_rubric_from_csv	Create a rubric from a CSV definition	confirm
associate_rubric	Attach a rubric to an assignment	confirm

Discussions & announcements

Tool	Purpose	confirm
list_discussion_topics	List a course's discussion topics
get_discussion_topic	Fetch one discussion topic
list_discussion_entries	List top-level entries in a topic
get_discussion_entry	Fetch one discussion entry
get_discussion_thread	Fetch a full thread with replies
post_discussion_entry	Post a new entry to a topic	confirm
reply_to_discussion_entry	Reply to an existing entry	confirm
create_discussion_topic	Create a discussion topic	confirm
update_discussion_topic	Update a discussion topic	confirm
list_announcements	List a course's announcements
create_announcement	Create an announcement	confirm
delete_announcement	Delete one announcement	confirm
delete_announcements	Delete multiple announcements by id	confirm
delete_announcements_matching	Delete announcements matching a filter	confirm

Modules

Tool	Purpose	confirm
list_modules	List a course's modules
list_module_items	List items in a module
create_module	Create a module	confirm
update_module	Update a module	confirm
delete_module	Delete a module	confirm
add_module_item	Add an item to a module	confirm
update_module_item	Update a module item	confirm
delete_module_item	Delete a module item	confirm

Pages

Tool	Purpose	confirm
list_pages	List a course's pages
get_page	Fetch one page's metadata
get_page_content	Fetch one page's body
get_front_page	Fetch a course's front page
create_page	Create a page	confirm
edit_page_content	Replace a page's body	confirm
update_page_settings	Update a page's settings (title, publish state, editing rights)	confirm
bulk_update_pages	Apply the same edit across multiple pages	confirm
delete_page	Delete a page	confirm

Files

Tool	Purpose	confirm
list_files	List a course's files
read_file	Read a text file's content
download_file	Download a file to disk
upload_file	Upload a file to a course	confirm

Messaging

Tool	Purpose	confirm
list_conversations	List Canvas Inbox conversations
get_conversation	Fetch one conversation's messages
get_unread_count	Get the unread message count
mark_conversations_read	Mark conversations as read	confirm
send_conversation	Send a new message	confirm
send_bulk_messages	Send the same message to multiple recipients	confirm

Peer reviews

Tool	Purpose	confirm
list_peer_reviews	List peer review assignments for an assignment
assign_peer_review	Assign a peer review	confirm
get_peer_review_assignments	List who is reviewing whom
get_peer_review_completion_analytics	Summarize peer review completion rates
get_peer_review_followup_list	List students who still owe a peer review
get_peer_review_comments	Fetch comments left in a peer review
analyze_peer_review_quality	Score peer review feedback for depth and usefulness
identify_problematic_peer_reviews	Flag low-effort or missing peer reviews
generate_peer_review_report	Build a summary report of peer review activity
generate_peer_review_feedback_report	Build a report of feedback content across peer reviews
extract_peer_review_dataset	Export peer review data for analysis
message_peer_reviewers	Message students about their peer review assignments	confirm
send_peer_review_followups	Send reminders to students with outstanding peer reviews	confirm

People & privacy

Tool	Purpose	confirm
list_users	List a course's enrolled users
list_groups	List a course's groups
check_enrollment	Check one user's enrollment in a course
export_anonymization_map	Export the pseudonym-to-real-name mapping	confirm
get_privacy_status	Show whether anonymization is active

Accessibility

Tool	Purpose	confirm
scan_course_content_accessibility	Scan a course's pages and files for accessibility issues
fix_accessibility_issues	Apply fixes for detected accessibility issues	confirm
fetch_ufixit_report	Fetch an existing UFIXIT accessibility report
parse_ufixit_violations	Parse violations out of a UFIXIT report
format_accessibility_summary	Format an accessibility scan as a readable summary

Course copy

Tool	Purpose	confirm
create_content_migration	Start a course content migration/copy	confirm
get_content_migration_status	Check a content migration's progress

Meta

Tool	Purpose	confirm
search_tools	Search the available tools by keyword

Configuration

All configuration comes from environment variables, loaded from a .env file in the repository root.

Variable	Required	Default	Purpose
CANVAS_URL	yes		Your Canvas host, e.g. https://yourschool.instructure.com. No /api/v1 suffix.
CANVAS_TOKEN	yes		Your Canvas personal access token.
CANVAS_TIMEOUT	no	30	HTTP request timeout, in seconds.
CANVAS_CACHE_TTL	no	300	How long course-resolution results are cached, in seconds.
CANVAS_MAX_CONCURRENCY	no	5	Maximum concurrent requests to Canvas.
CANVAS_ANONYMIZE_STUDENTS	no	false	Replace student names with stable pseudonyms in tool output.
CANVAS_DOWNLOAD_DIR	no	system temp directory	Where download_file writes downloaded files.

Run canvaslms-api --config to print the resolved configuration, with the token masked.

Development

Install with dev extras:

uv pip install -e ".[dev]"

Run tests:

pytest

Lint:

ruff check .

CI runs tests and lint on every push. A Docker image is also available; its entrypoint is canvaslms-api.

Useful CLI flags during development: canvaslms-api --list-tools prints every registered tool name, canvaslms-api --config prints resolved settings, canvaslms-api --version prints the installed version.

Security notes

• Your Canvas token stays local. It is read from .env on startup and used only to call the Canvas API you configured.
• The token is never logged. --config and get_privacy_status mask it.
• Write tools require an explicit confirm=true after a preview. Nothing changes in Canvas by accident.
• download_file will not overwrite an existing file at the destination path.

License

MIT. See LICENSE.