Canvas Lms MCP Local

[!TIP]

Want it on your phone, on the web, or across several computers? Go to the [Remote Edition →](https://github.com/Fireiscoolsolo/canvas-lms-mcp-server) (website). It has the same 237 tools, served over a password-protected tunnel with a status page, phone alerts and automatic failover between computers.

Canvas LMS MCP: Local Edition

Use Canvas from your AI assistant on your own computer. No tunnel, no passwords, no background service.

237 tools · One-command setup · Nothing exposed to the internet

Website · Quick start · Configuration · Tools · Troubleshooting · Changelog

[!TIP]

On Windows? Skip the terminal. The free [Canvas MCP Installer Assistant](https://fireiscoolsolo.github.io/canvas-lms-mcp-local/app/) sets this up in Claude Desktop, the ChatGPT desktop app and other AI apps with pictures and a check after every step, or does every step for you. [Download for Windows](https://fireiscoolsolo.github.io/canvas-lms-mcp-local/download/)

Local or Remote?

Highlights

• Your day and week: daily_briefing, get_week_ahead, semester_snapshot, cross-course missing work in one call
• Planner: add your own to-do notes, check items off your Canvas To Do list, export deadlines to Google, Apple or Outlook calendars
• Study plans: study_plan spreads study time before each deadline around your calendar; schedule_study_plan puts it on your Canvas calendar
• Grade intelligence: final_grade_scenarios (what you need for each letter, with group weights), what_if_grade, grade_impact_ranking, calculate_gpa
• Assignments: assignment_brief gives the instructions, rubric, late policy and linked files in one call
• Submissions, discussions and inbox: submit, post and message, always after you confirm (apps that support it ask you directly)
• Content: read lecture PDFs, Word docs and slides with read_course_file, plus pages, modules, captions and full course export
• Prompts and attachments: one-click prompts like *Daily briefing*, *Start this assignment* and *Am I on track to pass?*, and attach a syllabus or assignment to a chat

How it works

 Your AI app (MCP client)      ──stdio──►  node index.js  ──HTTPS──►  your school's Canvas API

(starts the server) (237 tools, runs only while your app is open)

No web server, no ports, nothing to keep running. Your AI app starts the server when it opens and stops it when it closes.

Quick start

[!NOTE]

Rather click than type? Download the [Canvas MCP Installer Assistant for Windows](https://fireiscoolsolo.github.io/canvas-lms-mcp-local/download/). It walks you through these steps with pictures and checks each one as you go, for both the Local and Remote editions. See screenshots.

Windows may show "Windows protected your PC" because the app isn't code-signed yet. Click More info, then Run anyway.

You need: Node.js 18+ (22+ to read PDFs) and a Canvas account.

1. Create a Canvas access token

In Canvas, go to Account → Settings → Approved Integrations → + New Access Token. Canvas only shows the token once, so copy it right away. Student tokens usually expire after 30–120 days.

2. Add it to your AI app

Easiest: one command (Windows, macOS and Linux)

npx canvas-lms-mcp-local setup

It asks for your school's Canvas address and your token, checks they work, then adds Canvas to the apps you pick: Claude Desktop, the ChatGPT desktop app, Claude Code, Cursor, VS Code, Windsurf, Gemini CLI and LM Studio. Your old settings are backed up first. It can also keep the token in Windows Credential Manager, the macOS Keychain or the Linux keyring instead of in each app's settings file.

Afterwards, fully quit and reopen your app (the command tells you how for each one).

Claude Desktop: one-click extension

1. Download `canvas-lms.mcpb` from the latest release.
2. Open it (double-click, or drag it into Claude Desktop's Settings → Extensions).
3. Click Install, then enter your school's Canvas address and your token when Claude asks.

The extension includes everything it needs, so you don't have to install Node.js for it.

Desktop apps, by hand

1. Open your app's MCP config file. It's usually under Settings → Developer or Settings → MCP servers, and most apps use an mcpServers JSON block. npx canvas-lms-mcp-local print-config prints the exact snippet for each app.
2. Add this canvas entry with your token and school address (also in mcp_config.example.json):

   {

"mcpServers": { "canvas": { "command": "npx", "args": ["-y", "canvas-lms-mcp-local"], "env": { "CANVAS_API_TOKEN": "your_token_here", "CANVAS_BASE_URL": "https://yourschool.instructure.com" } } } }

1. Fully quit the app: on Windows, right-click the tray icon and choose Quit; on macOS, press ⌘Q. Then reopen it. Closing the window isn't enough.

ChatGPT desktop app, by hand

1. Open Settings → Plugins, go to the MCPs tab, then click Add → Add MCP server.
2. Choose STDIO and fill in the form. On Windows, each argument goes in its own box:

| Environment variables | CANVAS_API_TOKEN = your token CANVAS_BASE_URL = https://yourschool.instructure.com |

1. Click Save, then start a new chat. Chats that were already open don't pick up new tools.

Command-line clients

Register this command as a stdio MCP server named canvas, with CANVAS_API_TOKEN and CANVAS_BASE_URL set in its environment:

npx -y canvas-lms-mcp-local

For Claude Code, npx canvas-lms-mcp-local setup --apps claude-code does it for you. Then check your client's MCP status to confirm it's connected.

3. Try it

Ask your assistant: "Run test_connection." It checks your identity, courses, planner, inbox and missing submissions, and reports any features your school has disabled.

Then try: *"Give me my daily briefing"*, *"What do I need on the rest of Biology to get a B?"*, *"Plan my study time for the next two weeks"* or *"Explain the Lab 4 assignment and its rubric."*

If something isn't working, run `npx canvas-lms-mcp-local doctor`. It checks Node, each app's settings and your token, and says how to fix what it finds.

Prefer running from source?

git clone https://github.com/Fireiscoolsolo/canvas-lms-mcp-local.git

cd canvas-lms-mcp-local npm install

Copy .env.example to .env in the same folder as `index.js`, fill in your token and school address, then point your config at the file instead of npx:

{ "mcpServers": { "canvas": { "command": "node", "args": ["C:/Users/YOUR-NAME/Documents/canvas-lms-mcp-local/index.js"] } } }

Forward slashes work on Windows. On macOS, the path looks like /Users/you/canvas-lms-mcp-local/index.js. On Windows, make sure the file is named .env, not .env.txt. To check, turn on *View → File name extensions* in File Explorer.

The tools live in core/, which is shared with the Remote Edition: change them here, then run npm run sync-core to copy them across. npm test runs every tool against a fake Canvas, so no token is needed.

Configuration

Put settings in the env block of your config, as in step 2. If you run from source, you can use .env next to index.js instead.

| `CANVAS_BASE_