mcp-classroom
MCP server for Google Classroom enabling reading assignments, checking submission states, uploading files to Drive, and viewing attachments, with optional Playwright automation for write operations restricted by Workspace admins.
README
mcp-classroom
MCP server for Google Classroom — read assignments, check submission states, upload files to Drive, and view attached files. Pairs with a Claude Code skill + Playwright for browser automation to work around Google Workspace write restrictions.
What it can do
| Feature | How |
|---|---|
| List courses | MCP |
| List & get assignments | MCP (slim response — no heavy file metadata) |
| Check submission state & grade | MCP |
| List files attached to a submission | MCP |
| Upload file to Google Drive | MCP |
| Attach file to submission | Playwright (API blocked by Workspace admins) |
| Turn in / Mark as done | Playwright (API blocked by Workspace admins) |
| Unsubmit (reclaim) | Playwright (API blocked by Workspace admins) |
Why Playwright? Google Workspace for Education admins can restrict third-party apps from modifying student submissions via the API (
@ProjectPermissionDenied). Playwright automates the browser instead, bypassing that restriction entirely.
Requirements
- Python 3.11+
- Node.js (for
playwright-cli) - A Google Cloud project with the Classroom API and Drive API enabled
1. Google Cloud Console setup
1.1 Create a project
- Go to console.cloud.google.com
- Create a new project (e.g.
mcp-classroom)
1.2 Enable APIs
In APIs & Services → Library, enable:
- Google Classroom API
- Google Drive API
1.3 Configure OAuth consent screen
- Go to APIs & Services → OAuth consent screen
- User type: External
- Fill in app name, support email, developer email
- Add scopes:
https://www.googleapis.com/auth/classroom.courses.readonly https://www.googleapis.com/auth/classroom.coursework.me https://www.googleapis.com/auth/classroom.coursework.students https://www.googleapis.com/auth/classroom.student-submissions.me.readonly https://www.googleapis.com/auth/classroom.rosters.readonly https://www.googleapis.com/auth/classroom.announcements.readonly https://www.googleapis.com/auth/drive.file https://www.googleapis.com/auth/drive.readonly - Add your Google account as a test user
- Publishing status: you can leave it in Testing while developing, or publish to Production to avoid the unverified app warning
1.4 Create OAuth credentials
- Go to APIs & Services → Credentials
- Click Create Credentials → OAuth client ID
- Application type: Desktop app
- Download the JSON file
- Rename it to
client_secret_<...>.jsonand place it in the project root (next topyproject.toml)
2. Install the MCP server
# Clone the repo
git clone https://github.com/your-username/mcp-classroom.git
cd mcp-classroom
# Install Python dependencies
pip install -e .
First run — authenticate
mcp-classroom
A browser window will open asking you to authorize the app with your Google account. After approval, a token.json is saved locally (gitignored).
3. Register with Claude Code
Add the server to your Claude Code MCP config (~/.claude/claude.json or via claude mcp add):
{
"mcpServers": {
"google-classroom": {
"command": "mcp-classroom",
"args": []
}
}
}
Or if running from source:
{
"mcpServers": {
"google-classroom": {
"command": "python",
"args": ["-m", "src.server"],
"cwd": "/absolute/path/to/mcp-classroom"
}
}
}
4. Install Playwright (for browser automation)
npm install -g playwright-cli
npx playwright install chromium
Save your Google session (one time)
playwright-cli open --browser=chrome https://accounts.google.com/signin
# Log in manually in the browser window
playwright-cli state-save ~/classroom-auth.json
playwright-cli close
5. Install the Claude Code skill (optional but recommended)
The skill tells Claude how to use the MCP + Playwright together without you having to explain the workflow every time.
- Copy
skill/SKILL.mdto your Claude skills directory:~/.claude/skills/google-classroom/SKILL.md - Edit the skill file and fill in your own course IDs and assignment IDs (run
list_coursesandlist_courseworkonce to get them). - Update the auth state path inside the skill to match where you saved
classroom-auth.json. - Reload plugins in Claude Code:
/reload-plugins
Available MCP tools
| Tool | Description |
|---|---|
list_courses |
List active courses (slim: id, name, section, state) |
get_course |
Get full details of a course |
list_coursework |
List assignments (slim: id, title, state, dueDate) |
get_coursework |
Get assignment details + description (no teacher file attachments) |
list_my_submissions |
List your submissions — state, grade, attachment count |
get_my_submission |
Get one submission's details |
list_submission_attachments |
List titles and Drive IDs of files you've attached |
upload_file_to_drive |
Upload a local file to Google Drive |
list_drive_files |
Search files in your Drive |
add_drive_attachment |
Attach a Drive file to a submission (may be blocked by Workspace) |
add_link_attachment |
Attach a URL to a submission (may be blocked by Workspace) |
remove_attachment |
Remove an attachment (may be blocked by Workspace) |
turn_in_submission |
Turn in a submission (may be blocked by Workspace) |
reclaim_submission |
Unsubmit a submission (may be blocked by Workspace) |
list_announcements |
List course announcements |
list_students |
List students in a course |
list_teachers |
List teachers in a course |
Tools marked "may be blocked by Workspace" return
403 @ProjectPermissionDeniedon institutional Google Workspace accounts. Use Playwright automation instead for those operations.
File structure
mcp-classroom/
├── src/
│ ├── server.py # MCP tool definitions
│ ├── classroom.py # Classroom API calls
│ ├── drive.py # Drive API calls
│ └── auth.py # OAuth flow + scopes
├── skill/
│ └── SKILL.md # Claude Code skill template
├── tests/
├── pyproject.toml
├── requirements.txt
└── README.md
Gitignored (never commit these)
token.json— your OAuth tokenclient_secret_*.json— your OAuth client credentials
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。