canvasMCP
Deploys a read-only MCP server to Cloudflare Workers for interacting with Canvas LMS, supporting courses, assignments, files, and more via a user-generated encrypted endpoint.
README
Canvas MCP on Cloudflare
This project deploys a read-only Canvas MCP server to Cloudflare Workers.
It includes a setup page so any user can:
- Enter their Canvas base URL
- Enter their Canvas access token
- Generate a private MCP endpoint to paste into Poke
The generated MCP URL contains an encrypted Canvas configuration. Treat it like a password.
What it supports
- Courses
- Assignments
- Files
- PDF text extraction
- Folders
- Modules
- Pages
- Announcements
- Discussion topics
- Quizzes
- Calendar events
- Todo items
Local development
- Install dependencies:
npm install
- Add a Worker secret for local
wrangler dev:
echo 'CONFIG_ENCRYPTION_SECRET=replace-with-a-long-random-string' > .dev.vars
- Start the Worker locally:
npm run dev
- Open the local URL shown by Wrangler, fill in the form, and copy the generated MCP URL.
Deploy to Cloudflare
- Authenticate if needed:
npx wrangler whoami
- Set the encryption secret:
npx wrangler secret put CONFIG_ENCRYPTION_SECRET
Use a long random value. Example:
openssl rand -base64 32
- Deploy:
npm run deploy
- Open the deployed Worker URL. Users can generate their own MCP URL from the setup page.
Poke setup
- In Canvas, create an access token from
Account > Settings. - Open the deployed setup page.
- Paste the Canvas base URL and token.
- Copy the generated MCP URL.
- In Poke, add an integration and use that generated MCP URL as the server URL.
Optional local Node mode
The original local server still works if you prefer fixed environment variables:
CANVAS_BASE_URL=https://canvas.school.edu \
CANVAS_TOKEN=your-token \
npm run start:http
License
Licensed under the GNU General Public License v3.0. See LICENSE.
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。