EduPage MCP Server

Project

A Model Context Protocol (MCP) server that exposes the full functionality of the edupage-api Python library to AI agents such as opencode, Claude, Cursor and any other MCP client.

EduPage is a school information system used across Europe. This server lets you query and operate a student / teacher / parent EduPage account directly from your agent: timetables, grades, homework, substitutions, meals (including ordering), messages, rosters, parent child-switching and more — including multiple schools (e.g. two children attending different schools).

⚠️ Unofficial API. Like all EduPage MCP servers, this relies on the community-maintained edupage-api library, which talks to EduPage's undocumented endpoints. Use read-only features freely; use the write features (send_message, meal ordering, child switching) carefully.

Table of Contents

• Why another EduPage MCP server?
• What it provides
• Getting started
  • Prerequisites
  • 1. Install
  • 2. Configure credentials
  • 3. Register with your MCP client
• Prompt examples
• Multiple schools (subdomains)
• Tool reference
• Data & safety notes
• Contributing
• Limitations
• Support
• License

Why another EduPage MCP server?

Two other EduPage MCP servers already exist:

• mrtineu/edupage-mcp — also published on PyPI as edupage-mcp
• mhlavac/edupage-mcp

Both are good and I have no affiliation with them — they are simply referenced here for honest comparison. They primarily focus on the read-only surface of the API.

This project deliberately goes further:

Key differentiators:

• Multi-school automatic discovery. Set EDUPAGE_SUBDOMAINS with one shared login and the server auto-discovers students across all schools — no need to maintain a manual "Student → school1, Student → school2" mapping. A student at two schools (e.g. Student at school1 + school2) is found automatically with separate per-school results.
• Role-aware tools. The server detects whether you're a parent, student, or teacher at each school and behaves accordingly — get_student_timetable switches to the student account for parents, returns direct timetables for students. No tool duplication.
• Full write surface. Meal ordering/rating, message sending, student switching — the other servers don't cover these.

Comparison checked 2026-09-28. If a row is wrong, please open an issue.

What it provides

A single stdio MCP server exposing 31 tools (published on PyPI as edupage-mcp-full):

• Authentication — login (by credentials, portal auto-detect, or a PHPSESSID cookie via method=), login_all (multi-school, one call), two_factor_finish (complete a pending 2FA), get_subdomains (schools available to the account with role/user id per school + env config, live subdomain discovery for parents). The former auth_status, user_id, login_auto, login_from_session, and two_factor_check_confirmed tools are folded into these.
• Timetables — get_my_timetable, get_timetable (teacher/student/class/ classroom; end_date for a range, formerly get_timetable_range), get_student_timetable (student by name, cross-school), get_next_week_timetable, get_next_ringing_time, get_periods, get_school_year
• Students — find_student (name → person_id, cross-school), get_student_timetable (cross-school, role-aware), scan_students (auto-discover all students across schools), get_my_students (classmates or school-wide for parents), switch_to_student (by id or name, parent only), switch_to_parent, clear_student_cache (force refresh cached student lists)
• Schools — get_subdomains (subdomains available to the account — live-discovered for parents, limited by EDUPAGE_SUBDOMAINS when set — plus session state per school)
• Grades — get_grades
• Timeline / notifications — get_timeline (category= for homework, assignments, absences, events, news, or full history since a date)
• Homework — get_timeline(category='homework') to list assignments (ids in additional_data.superid), get_homework_material to read one assignment's body text, dates and attachment list (the notification alone only carries the title and due date), download_attachment to save one attachment to disk
• Substitutions — get_timetable_changes, get_missing_teachers
• Meals — get_meals, choose_meal, sign_off_meal, rate_meal
• Day summaries — get_day_summary (one call: timetable, substitutions, missing teachers, grades, meals, homework, assignments, absences, news, events, notifications for a date — "what happened yesterday at school" in a single round trip; each section is isolated so one failure doesn't kill the report). Includes an OpenCode skill (school-day-summary) for human-readable formatting in OpenCode; other clients use the raw JSON directly.
• Rosters — get_roster (roster_type= for students, all students, teachers, classes, classrooms, or subjects), get_my_students
• Actions — send_message, switch_to_student, switch_to_parent, custom_request

Getting started

You need an MCP-capable client (opencode, Claude Desktop, Cursor, etc.).

1. Install

If you are using an AI coding client, a simple prompt is often enough to get started, for example: "Install the EduPage MCP as described in this GitHub repository oliverhruby/edupage-mcp". Most MCP-capable clients can then guide you through the available setup options.

Option A — from MCP Registry (recommended, one-click in VS Code / GitHub Copilot)

The server is listed in the MCP Registry. In VS Code or GitHub Copilot, search for "EduPage MCP" and install with one click. Or use the direct deeplink: mcp://install/io.github.oliverhruby/edupage-mcp

Option B — from PyPI

Use this for normal usage with a released version.

Requirements: uv for uvx, or Python 3.10+ for pip.

uvx edupage-mcp-full
# or, if you prefer pip (into whatever environment your MCP client uses):
pip install edupage-mcp-full

uvx runs the package without a persistent install. If uvx is unavailable, install uv first (pip install uv or winget install astral-sh.uv).

Option C — from GitHub (latest source)

Use this if you want the latest changes before a PyPI release.

Requirements: uv for uvx, or Python 3.10+ for pip.

uvx --from "git+https://github.com/oliverhruby/edupage-mcp.git" edupage-mcp-full
# or
pip install "git+https://github.com/oliverhruby/edupage-mcp.git"

Option D — Docker

Use this for an isolated container runtime.

Requirements: Docker.

Pull a prebuilt image (recommended):

docker pull ghcr.io/oliverhruby/edupage-mcp:latest

docker run --rm -i \
  -e EDUPAGE_USERNAME=your_username \
  -e EDUPAGE_PASSWORD=your_password \
  ghcr.io/oliverhruby/edupage-mcp:latest

Version tags are also available (for example v0.4.0) if you prefer pinned images.

Build locally from source (fallback):

docker build -t edupage-mcp-full .

docker run --rm -i \
  -e EDUPAGE_USERNAME=your_username \
  -e EDUPAGE_PASSWORD=your_password \
  edupage-mcp-full

The container uses the same environment variables described in Configure credentials. It also includes a HEALTHCHECK (stdio process liveness by default; local TCP check in HTTP transport modes).

For HTTP transports, set optional runtime vars:

• MCP_TRANSPORT: stdio (default), sse, or streamable-http
• MCP_HOST: bind host (default 127.0.0.1)
• MCP_PORT: bind port (default 8000)
• MCP_API_KEY: optional bearer token for HTTP auth

When MCP_API_KEY is set, HTTP requests must include Authorization: Bearer <key>. If MCP_API_KEY is not set, HTTP endpoints are unauthenticated. For production, prefer proper authentication and TLS via a reverse proxy or API gateway.

pyproject.toml pins mcp<2 (the stable FastMCP v1 API). mcp 2.x renamed FastMCP to MCPServer and changed the API surface; this server targets the FastMCP v1 API for simplicity and stability.

Option C — development from source

Use this if you are contributing or debugging locally.

Requirements: Python 3.10+.

git clone https://github.com/oliverhruby/edupage-mcp.git
cd edupage-mcp
uv sync               # or: python -m venv .venv && .venv/bin/python -m pip install -e .
uv run edupage-mcp-full

Option D — Docker

Use this for an isolated container runtime.

Requirements: Docker.

Pull a prebuilt image (recommended):

docker pull ghcr.io/oliverhruby/edupage-mcp:latest

docker run --rm -i \
  -e EDUPAGE_USERNAME=your_username \
  -e EDUPAGE_PASSWORD=your_password \
  ghcr.io/oliverhruby/edupage-mcp:latest

Version tags are also available (for example v0.4.0) if you prefer pinned images.

Build locally from source (fallback):

docker build -t edupage-mcp-full .

docker run --rm -i \
  -e EDUPAGE_USERNAME=your_username \
  -e EDUPAGE_PASSWORD=your_password \
  edupage-mcp-full

The container uses the same environment variables described in Configure credentials. It also includes a HEALTHCHECK (stdio process liveness by default; local TCP check in HTTP transport modes).

For HTTP transports, set optional runtime vars:

• MCP_TRANSPORT: stdio (default), sse, or streamable-http
• MCP_HOST: bind host (default 127.0.0.1)
• MCP_PORT: bind port (default 8000)
• MCP_API_KEY: optional bearer token for HTTP auth

When MCP_API_KEY is set, HTTP requests must include Authorization: Bearer <key>. If MCP_API_KEY is not set, HTTP endpoints are unauthenticated. For production, prefer proper authentication and TLS via a reverse proxy or API gateway.

pyproject.toml pins mcp<2 (the stable FastMCP v1 API). mcp 2.x renamed FastMCP to MCPServer and changed the API surface; this server targets the FastMCP v1 API for simplicity and stability.

2. Configure credentials

Either set environment variables or pass credentials to login (see Prompt examples).

# Windows (persistent, per-user)
setx EDUPAGE_USERNAME "your_username"
setx EDUPAGE_PASSWORD "your_password"
setx EDUPAGE_SUBDOMAINS "s1,s2,s3"       # optional: multiple schools (auto-login + discovery)

# macOS / Linux
export EDUPAGE_USERNAME="your_username"
export EDUPAGE_PASSWORD="your_password"
export EDUPAGE_SUBDOMAINS="s1,s2,s3"     # optional

Single school? Just set EDUPAGE_USERNAME + EDUPAGE_PASSWORD. The server auto-discovers your school via the EduPage portal on startup — no subdomain needed.

Multiple schools? Add EDUPAGE_SUBDOMAINS (comma-separated). The server logs into all of them on startup with your shared credentials.

3. Register with your MCP client

opencode — add to ~/.config/opencode/opencode.json (or opencode.jsonc):

{
  "mcp": {
    "edupage": {
      "type": "local",
      "enabled": true,
      "command": ["uvx", "edupage-mcp-full"],
      "env": {
        "EDUPAGE_USERNAME": "{env:EDUPAGE_USERNAME}",
        "EDUPAGE_PASSWORD": "{env:EDUPAGE_PASSWORD}",
        "EDUPAGE_SUBDOMAINS": "{env:EDUPAGE_SUBDOMAINS}"
      }
    }
  }
}

Put credentials in your shell/environment (or a .env) and reference them with {env:VAR}, or hardcode them under env: directly. uvx will auto-provision the package the first time; it must be on your PATH.

Claude Desktop / Cursor — use claude_desktop_config.json / .mcp.json with a mcpServers entry in the standard shape, pointing command/args at the venv python and the edupage_mcp.py path, plus an env block with your credentials.

After editing client config, restart the client so the MCP server is loaded.

Prompt examples

Multiple schools & automatic student discovery

Each subdomain (school) keeps its own logged-in session. There are two ways to log in to several schools at once:

A) Automatic on startup (recommended). Set EDUPAGE_SUBDOMAINS (a comma-separated list) plus the shared EDUPAGE_USERNAME / EDUPAGE_PASSWORD — the server logs into all of them when it launches, so every tool is immediately ready and students are discoverable across all schools with no login call and no student→school mapping:

setx EDUPAGE_SUBDOMAINS "school1,school2,school3"   # Windows
export EDUPAGE_SUBDOMAINS="school1,school2,school3" # macOS / Linux

get_subdomains   # lists school1, school2, school3 (logged in, with role)
scan_students      # discovers Student A and Student B across those schools
get_student_timetable name="Student A"   # is found at school1 AND school2

B) On demand with login_all. Authenticate several schools at once, then pass subdomain to any data tool (it defaults to the last active subdomain when omitted):

login_all subdomains="school1,school2" usernames="u1,u2" passwords="p1,p2"

get_my_timetable subdomain="school1"
get_my_timetable subdomain="school2"
get_subdomains         # shows all logged-in subdomains + which is active

You can also call login once per school to add/lookup sessions incrementally.

Single school? No EDUPAGE_SUBDOMAINS needed — the server auto-discovers your school via the portal on startup. For two or more schools, set EDUPAGE_SUBDOMAINS (auto-login) or use login_all / repeated login calls.

Students by name (e.g. "timetable for Student A")

Because the server auto-discovers students across the configured EDUPAGE_SUBDOMAINS (or every logged-in school when the variable is unset), you don't need to know or state which school a student is in. Just ask for the timetable by name and the server searches every school in scope:

"timetable for Student A"  ->  get_student_timetable name="Student A"

get_student_timetable (with no subdomain):

1. searches every school in the discovery scope — the configured EDUPAGE_SUBDOMAINS, or all logged-in schools when unset — for a student whose first/last/full name matches (scan_students does just the discovery step),
2. for each school where the student is found, switches to the student account if you're logged in as a parent, returns that student's timetable for the date, and switches back to the parent account afterwards,
3. returns one result per school.

A student attending more than one school (e.g. Student at school1 + school2) therefore yields a list of two per-school timetables — separate results, never merged. This is the built-in replacement for maintaining a manual "Student → school1" mapping: with EDUPAGE_SUBDOMAINS set, discovery is fully automatic.

Tool reference

Data & safety notes

• Most tools are read-only. The ones marked Writes? ✅ mutate EduPage state (sent messages, ordered meals, switched accounts). Use them with care.
• download_attachment is the one tool that writes to local disk: it saves into dest_dir (default <tempdir>/homework), takes the name from the server's content-disposition when present, and appends (1), (2), ... rather than overwriting anything. An HTML login/error page is refused instead of being written out as an attachment.
• custom_request returns text only: a binary body is refused with a pointer to download_attachment, and a text body over 60000 characters comes back truncated with an explicit truncated: true.
• get_timeline categories homework, assignments, absences, events and news derive their data from the timeline notifications — if the school doesn't push certain event types, those categories may return empty lists.
• get_missing_teachers is marked experimental upstream (parses HTML from the substitution page) and can raise if a teacher's name no longer matches.
• Meal rate_meal and ordering depend on the school publishing menus with the matching identifiers; not all schools expose ratings.
• get_meals first tries the per-student meal-ordering endpoint (needed for ordering/ratings). When a school doesn't enable that, it falls back to the school's public canteen menu widget (/menu/?wid=menu_CanteenMenu_1). All five slots (breakfast/snack/lunch/afternoon_snack/dinner) are always returned; slots the school doesn't publish are None.

Skills

The package includes an OpenCode skill (school-day-summary) at <site-packages>/edupage_mcp/skills/school-day-summary/SKILL.md. It teaches OpenCode agents how to turn get_day_summary JSON into a human-readable daily school report.

OpenCode only: To register it:

mkdir -p ~/.config/opencode/skills/school-day-summary
cp <site-packages>/edupage_mcp/skills/school-day-summary/SKILL.md \
   ~/.config/opencode/skills/school-day-summary/SKILL.md

Restart OpenCode; the agent can then answer "what happened at school yesterday for my kids?" by calling get_day_summary per child.

Other MCP clients (Copilot, Claude, Cursor, etc.) — call get_day_summary directly; they receive the full structured JSON. Formatting is client-specific (no skill system in the MCP protocol).

Contributing

Contributor and maintainer guidance is in CONTRIBUTING.md.

• Contribution workflow and local setup
• Architecture and implementation details
• Release process (PyPI, GitHub Releases, GHCR)
• CI quality gates and upstream coverage drift checks

Limitations

• Unofficial/read-mostly by design. EduPage can change its endpoints at any time; reliability ultimately depends on edupage-api, not this wrapper.
• No CAPTCHA bypass. If EduPage presents a CAPTCHA during login, log in via browser first, then use login method="session" with the resulting PHPSESSID.
• 2FA requires human interaction (approve on device or provide a code).
• Parent/teacher accounts are only partially verified upstream; some parent methods are best-effort.
• The auth session lives for the lifetime of the MCP server process; restarting the client means logging in again.
• Cross-school student discovery depends on being logged into all relevant schools (via EDUPAGE_SUBDOMAINS, login_all, or repeated login calls). If a school is not logged in, that student's results from that school cannot be discovered.

Support

If you like this project and want to support or request a feature, send me a beer, it keeps my mind relaxed and ideas will come :-)

License

MIT © Oliver Hrubý

This project is not affiliated with or endorsed by Ascora (EduPage) or by the authors of edupage-api. EduPage is a registered trademark of its respective owner(s).