ParentSquare MCP
MCP server that gives Claude read access to your ParentSquare parent account, enabling queries about school posts, calendars, messages, directory, signups, forms, payments, and files. It also parses image/PDF attachments so calendar flyers are accessible.
README
ParentSquare MCP
An MCP server that gives Claude read access to your own ParentSquare parent account — school posts, calendars, messages, the staff directory, signups, forms, payments and files.
School calendars and flyers are very often posted as image or PDF attachments rather than as calendar entries. This server downloads them and hands them to Claude as real image/text content, so "what's on the school calendar next week?" works even when the calendar itself is empty.
Before you start: how this connects
ParentSquare has no public parent-facing API. Their documented API is district-side, for pushing student data in from an SIS — it is issued to district IT, not to parents, and it is not what you want here. So this server signs in as you and reads the same parent web app you use in a browser.
Two consequences worth knowing up front:
- You need to supply your own session. See Authentication.
- The page layouts are not a contract. Districts run different versions of the app on different
URLs, so there are three layers of defence: every route is a list of candidate paths tried in
order; if all of them fail, the section is located by matching its label in your sidebar; and
if a parser still finds nothing, it returns the readable page text rather than failing.
scripts/doctor.pyreports what needed which. See Fitting it to your district.
Install
cd "~/Desktop/ParentSquare MCP"
python3 -m venv .venv
./.venv/bin/python -m pip install -e .
Authentication
Method 1 — browser cookie (recommended)
Works with every district, including ones that sign in through Google, Clever or ClassLink, and with two-factor accounts.
- Sign in to ParentSquare in your browser.
- Open DevTools → Network, click any request to
parentsquare.com. - Under Request Headers, copy the entire value of
Cookie:. - Put it in
PARENTSQUARE_COOKIE(see Connect it to Claude).
The cookie expires eventually — when tools start reporting that the session expired, repeat this.
Method 2 — email and password
Only works if your district uses a plain ParentSquare password, with no two-factor.
PARENTSQUARE_EMAIL=you@example.com ./.venv/bin/python scripts/login.py
This caches the session in ~/.parentsquare-mcp/session.json (mode 600). If your district uses SSO
or 2FA, the script says so and points you back to Method 1.
Connect it to Claude
Claude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"parentsquare": {
"command": "/Users/chen/Desktop/ParentSquare MCP/.venv/bin/python",
"args": ["-m", "parentsquare_mcp"],
"cwd": "/Users/chen/Desktop/ParentSquare MCP",
"env": {
"PARENTSQUARE_COOKIE": "_parentsquare_session=…; remember_user_token=…"
}
}
}
}
Restart Claude Desktop.
Claude Code
claude mcp add parentsquare \
--env PARENTSQUARE_COOKIE="_parentsquare_session=…" \
-- "/Users/chen/Desktop/ParentSquare MCP/.venv/bin/python" -m parentsquare_mcp
If you used scripts/login.py, the cached session is picked up automatically and you can drop the
env block entirely.
See .env.example for every setting.
Fitting it to your district
Run this once after connecting:
./.venv/bin/python scripts/doctor.py
It reports, for every section, which URL works and how many rows the parser got:
— routes —
ok feed /feeds
ok directory /directory
NAV payments /school_pay (found via the sidebar)
MISS volunteer_hours (no candidate worked, and no sidebar link matched)
— tools (23 registered) —
ok get_feeds 12 rows
THIN list_signups 0 rows — selectors need work. Source: https://…/signups
- NAV — none of the guessed URLs worked, so the server found the section by matching its
sidebar label instead. This already works; adding the reported path to
routes.pyjust saves a lookup. Districts re-route these pages constantly, which is why label matching is the safety net. - MISS — no URL worked and no sidebar link matched. Open the section in your browser, copy the
path from the address bar, and add it to the front of the matching list in
parentsquare_mcp/routes.py. If the sidebar calls it something unusual, adding that word toSECTION_KEYWORDSin the same file fixes it for good. - THIN — the page loaded but nothing matched. Ask Claude to call
debug_fetchon that URL to see the real markup, then adjust the selector list at the top of the relevant file inparentsquare_mcp/tools/. The selectors are plain CSS lists, deliberately easy to edit. - empty — the page loaded and genuinely has nothing in it. Nothing to fix.
Nothing in the doctor writes to ParentSquare.
Tools
Feed and posts
| Tool | What it does |
|---|---|
get_feeds |
Paginated feed: titles, authors, dates, summaries, attachment names. Takes a query filter. |
get_post |
Full post — body, comments, poll results, signup items, and the contents of attached images and PDFs. |
get_group_feed |
Posts from one group. |
Calendar
| Tool | What it does |
|---|---|
get_calendar_events |
Structured events from the ICS feed, recurring events expanded. When empty, explains how to find calendars posted as attachments. |
Communication
| Tool | What it does |
|---|---|
list_conversations / get_conversation |
Read message threads. |
get_directory |
Staff directory: name, role, phone, email, user id. |
get_staff_member |
Full profile including office hours and profile photo. |
Media and files
| Tool | What it does |
|---|---|
list_photos |
Photo gallery with image URLs. |
list_files |
Shared documents with download URLs. |
download_file |
Save any attachment to disk. |
Things to act on
| Tool | What it does |
|---|---|
list_signups |
Sign-ups and RSVPs with progress (e.g. 53/103 Items). |
list_notices |
Alerts and secure documents. |
list_polls |
Polls with vote counts and the leading option. |
list_forms |
Permission slips, with an outstanding count. |
list_payments |
Payment items with prices and totals. |
list_volunteer_hours |
Logged hours with the total. |
Discovery
| Tool | What it does |
|---|---|
list_schools |
Schools and students, with the ids other tools take. |
list_school_features |
Which sections a school actually has, from its sidebar. |
list_groups |
Groups with member counts and membership status. |
list_links |
Pinned quick links. |
get_student_dashboard |
A student's school, grade, classes and teachers. |
debug_fetch |
Escape hatch: fetch any ParentSquare path and see what is really there. |
Nothing here writes to ParentSquare — every tool is read-only against your account. On your own disk,
download_file saves where you ask it to, and get_post saves image-only PDFs (scans, which have no
extractable text) to ~/Downloads/ParentSquare so they are still reachable.
Things to try
What did the school send home this week?
Is there anything I still need to sign or pay for?
The October calendar was posted as a PDF — what days is my kid off?
Who is my son's teacher and what's their email?
Tests
./.venv/bin/python tests/run_all.py
Four suites — parsers, tools over HTTP, a real MCP handshake over stdio, and the authentication paths — all against a fake ParentSquare served on localhost. They never touch the real site, so they are safe to run at any time.
Layout
parentsquare_mcp/
config.py environment settings
client.py auth, HTTP, route fallback, login-bounce detection
routes.py candidate URLs + sidebar keywords ← edit when doctor says MISS
parsers.py shared HTML helpers and the fallback result shape
media.py attachments → images / PDF text
tools/ one module per group of tools ← edit when doctor says THIN
scripts/
login.py cache a session with email + password
doctor.py check routes and tools against your account
tests/
Notes
- Read-only by design. Nothing posts, replies, signs, pays or RSVPs on your behalf.
- Your credentials stay on your machine: in the client config env block, or in
~/.parentsquare-mcp/session.jsonat mode 600. - Because it depends on the parent web app's HTML, a ParentSquare redesign can break parsing. That is
what
doctor.pyanddebug_fetchare for, and why nothing hard-fails when a selector misses.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.