schoology-mcp
An MCP server that automates login to PAUSD Schoology via ClassLink and provides tools to fetch grades, upcoming assignments, and recent posts.
README
Schoology MCP
An MCP server for PAUSD Schoology (https://pausd.schoology.com). It drives
a headless browser that logs in automatically through the ClassLink portal
and exposes tools to fetch grades, upcoming assignments and recent posts.
It builds on the scraping approach from dajun666/schoology-get (Playwright +
BeautifulSoup), and adds the missing piece: automated login — no more
hand-exporting cookies.json.
How login works
PAUSD students reach Schoology through ClassLink. The server:
- Opens
https://launchpad.classlink.com/pausdand fills the student ID + password. - Clicks the Schoology tile, which performs SAML SSO into Schoology.
- Saves the session to
storage_state.jsonso later runs skip the login until it expires (then it logs in again automatically).
Setup
python3 -m venv .venv
source .venv/bin/activate # do this in every new shell
pip install -r requirements.txt
playwright install chromium
cp .env.example .env
# edit .env: set SCHOOLOGY_USERNAME (your 8-digit student ID)
# store the password in the OS keychain -- not in any file:
python scripts/set_credentials.py
The python … commands below assume this .venv is activated. The MCP server
itself runs outside any shell, so it is registered with the venv's Python by
absolute path — see Register with Claude Code.
Credentials & password storage
The password is never stored in a plaintext file. set_credentials.py
saves it to the OS keychain (macOS Keychain) via the keyring library —
encrypted at rest and unlocked by your macOS login. The server reads it from
there at runtime.
- Update it later: re-run
python scripts/set_credentials.py. - Remove it:
python scripts/set_credentials.py --delete. - macOS may ask once to allow access to the keychain item — choose Always Allow so the unattended server isn't blocked.
SCHOOLOGY_USERNAME(the student ID) stays in.env; it is the keychain lookup key, not a secret.- Fallback: if you set the
SCHOOLOGY_PASSWORDenvironment variable, it is used instead of the keychain (handy for throwaway/CI use).
Caveat: the keychain protects the password at rest and keeps it out of dotfiles,
backups and git — but any process running as your macOS user can still read it.
It is strictly better than a plaintext .env, not a sandbox.
Verify login (do this first)
python scripts/login_check.py --show-browser
This logs in, saves storage_state.json, and dumps page HTML into dumps/.
All four parsers are verified against a real PAUSD account (grades, courses,
upcoming assignments, recent posts). If Schoology changes its markup later,
re-run this and re-check the selectors in schoology_mcp/parsers.py against
the fresh dumps/.
Tools
| Tool | Description |
|---|---|
get_grades |
Current grades for every course (periods, categories, assignments). |
get_courses |
Enrolled courses. |
get_upcoming_assignments |
Upcoming / due-soon assignments and events. Pass include_info=True to also fetch each item's description (slower). |
get_assignment_info |
Full details (title, course, due, description, attachments) of one assignment. Takes a URL, /assignment/NNN path, or bare id. |
get_recent_posts |
Latest posts from the activity feed. |
Run
python server.py # speaks MCP over stdio
Test interactively with the MCP Inspector:
npx @modelcontextprotocol/inspector python server.py
Register with Claude Code
claude mcp add schoology -- /your/path/to/schoology-mcp/.venv/bin/python /your/path/to/schoology-mcp/server.py
Or add it to an MCP client config (e.g. Claude Desktop claude_desktop_config.json):
{
"mcpServers": {
"schoology": {
"command": "/your/path/to/schoology-mcp/.venv/bin/python",
"args": ["/your/path/to/schoology-mcp/server.py"],
"env": {
"SCHOOLOGY_USERNAME": "950XXXXX"
}
}
}
}
No password appears in the config — it is read from the OS keychain (set once
via scripts/set_credentials.py). SCHOOLOGY_USERNAME can come from .env or
the env block above.
Sessions & auto-refresh
Schoology sessions expire quickly. The server handles this automatically:
- Detect-and-retry — every tool call checks (by page content, not just URL) whether the page it got back is really logged in. If the session died, it re-logs in via ClassLink and retries — a call never silently returns logged-out data.
- Keep-alive — a background task re-visits Schoology every
SCHOOLOGY_KEEPALIVE_MINUTES(default 8) to keep the session warm, so interactive calls rarely wait for a fresh login. - Persistence — the refreshed session is saved to
storage_state.json, so restarting the server reuses it instead of logging in again.
Set SCHOOLOGY_KEEPALIVE=false to disable the background task; detect-and-retry
still applies.
Notes
.envandstorage_state.jsonhold credentials/session — they are git-ignored. Never commit them.- Set
SCHOOLOGY_HEADLESS=falsein.envto watch the browser while debugging.
Contributing
Feature requests and issues are always welcomed — open one on the issue tracker or send a PR.
Forking for your own district
This repo is hard-wired to PAUSD (pausd.schoology.com + the ClassLink
/pausd tenant). It is intentionally easy to retarget: fork the repo and
edit the two URLs (and, if your district doesn't use ClassLink SSO,
schoology_mcp/auth.py).
| What to change | Where |
|---|---|
| Schoology base URL | SCHOOLOGY_BASE_URL in .env (or the default in schoology_mcp/config.py) |
| ClassLink tenant URL | CLASSLINK_URL in .env (e.g. https://launchpad.classlink.com/<your-district>) |
| Login flow (if not ClassLink) | schoology_mcp/auth.py — replace the ClassLink portal step with your district's IdP (Clever, Google SSO, direct Schoology login, etc.). The Schoology-side scraping in schoology_mcp/parsers.py is district-agnostic and should keep working. |
If you ship a working fork for another district, open an issue with a link — we can list known-good forks here.
License
MIT — fork it, ship it, no warranty.
Star History
<a href="https://www.star-history.com/?repos=dajun666%2Fschoology-mcp&type=date&legend=top-left"> <picture> <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/chart?repos=dajun666/schoology-mcp&type=date&theme=dark&legend=top-left" /> <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/chart?repos=dajun666/schoology-mcp&type=date&legend=top-left" /> <img alt="Star History Chart" src="https://api.star-history.com/chart?repos=dajun666/schoology-mcp&type=date&legend=top-left" /> </picture> </a>
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。