penn-course-mcp
Enables course planning for University of Pennsylvania by searching the catalog, inspecting sections and ratings, detecting schedule conflicts, and building weekly schedules.
README
penn-course-mcp
An MCP server for University of Pennsylvania course planning. It is built on the public Penn Courses API (the same backend behind Penn Course Plan and Penn Course Review), so the catalog, ratings, and schedule data are live.
With it, an MCP client (Claude Desktop, Claude Code, or any other) can search the course catalog, inspect sections and meeting times, read ratings and reviews, check a set of sections for time conflicts, assemble a weekly schedule, and save that schedule back to a Penn Course Plan account.
The server works with no configuration against the public API. Setting a session cookie additionally unlocks detailed Penn Course Review breakdowns and the schedule-write tools. It runs over stdio for local clients, or streamable HTTP when hosted.
Tools
| Tool | What it does |
|---|---|
get_current_semester |
Active term code (e.g. 2026C) and review-auth status |
search_courses |
Search the catalog by code or keyword; returns summaries and ratings |
get_course_details |
Full course info: description, prereqs, attributes, sections, meeting times |
list_course_sections |
Sections with formatted meeting times and instructors; filter by status or activity |
get_course_ratings |
Public aggregate ratings (quality, difficulty, workload) |
get_course_reviews |
Detailed PCR reviews when authenticated; falls back to aggregates otherwise |
find_courses_by_attribute |
Courses carrying a given attribute or Gen-Ed code |
find_courses_by_requirement |
Courses fulfilling a requirement (by code or name) |
check_schedule_conflicts |
Detect time conflicts among a set of section ids |
build_schedule |
Weekly grid, total credits, conflicts, and missing-companion warnings |
compare_courses |
Side-by-side ratings, difficulty, workload, and prereqs |
recommend_courses |
Search and rank by quality, difficulty, or workload |
list_schedules |
List the saved schedules in your Penn Course Plan account (auth) |
save_schedule |
Write a schedule to your Penn Course Plan account so it shows on the site (auth) |
delete_schedule |
Delete a saved Penn Course Plan schedule by id (auth) |
Course codes look like CIS-1200; section ids look like CIS-1200-001.
Installation
uv pip install -e ".[dev]" # or: pip install -e ".[dev]"
Running
# stdio (default), for Claude Desktop / Claude Code
penn-course-mcp
# streamable HTTP at http://127.0.0.1:8000/mcp
penn-course-mcp --transport http --port 8000
# development server / MCP Inspector
fastmcp dev src/penn_course_mcp/server.py
Client configuration
Add the server to your MCP client config (for Claude Desktop, this is
claude_desktop_config.json):
{
"mcpServers": {
"penn-course": {
"command": "uvx",
"args": ["penn-course-mcp"],
"env": {
"PENN_COURSES_SESSION_COOKIE": ""
}
}
}
}
Use
"command": "penn-course-mcp"instead ofuvxif you installed the package into a virtualenv that is on yourPATH.
Configuration
All environment variables are optional (see .env.example):
| Variable | Default | Purpose |
|---|---|---|
PENN_COURSES_BASE_URL |
https://penncoursereview.com |
API base URL |
PENN_COURSES_SEMESTER |
current |
Default semester (current auto-resolves) |
PENN_COURSES_CACHE_TTL |
3600 |
Cache TTL (seconds) for catalog and detail data |
PENN_COURSES_TIMEOUT |
20.0 |
HTTP timeout (seconds) |
PENN_COURSES_USER_AGENT |
penn-course-mcp/<ver> |
Sent with every request |
PENN_COURSES_SESSION_COOKIE |
(unset) | Enables detailed reviews and schedule writes (see below) |
PENN_COURSES_TRANSPORT / --transport |
stdio |
stdio or http |
PENN_COURSES_HOST / --host |
127.0.0.1 |
HTTP host |
PENN_COURSES_PORT / --port |
8000 |
HTTP port |
Detailed reviews and schedule writes
The public API exposes aggregate ratings (course and instructor quality, difficulty, and work required) without authentication. The detailed per-instructor Penn Course Review breakdowns, and the schedule-write tools, require a logged-in Penn session. To enable them, log in to penncoursereview.com, copy your session cookie, and set:
export PENN_COURSES_SESSION_COOKIE="sessionid=...; csrftoken=..."
get_course_reviews falls back to the public aggregate ratings when no cookie is set, so it
never errors. The schedule-write tools (list_schedules, save_schedule, delete_schedule)
read and modify a real Penn Course Plan account, so they always need the cookie. Writes also
need the csrftoken value, since Django enforces CSRF on unsafe methods. save_schedule
creates a new schedule by default; pass an existing schedule_id (from list_schedules) to
overwrite one instead.
The read-only catalog and rating tools never modify your account. Only the three schedule tools do.
Claude skill
This repository ships an optional Agent Skill
at .claude/skills/penn-course-planning. It
gives Claude guidance on using the tools well, including normalizing course codes (CIS 120
to CIS-1200), reading the rating scales, and choosing the right tool for schedule and
requirement questions. It activates automatically when you use Claude Code in this repository.
To make it available everywhere, copy it into your user skills folder:
cp -R .claude/skills/penn-course-planning ~/.claude/skills/
Development
uv pip install -e ".[dev]"
pytest # pure planning logic, mocked client, and tool tests
flake8 # style, import order (isort), and quote checks
black . # formatter
The planning logic in src/penn_course_mcp/planning.py is pure and network-free, so the
schedule-conflict detection, time formatting, and comparison are covered by unit tests
without hitting the API.
Notes
- Meeting times use Penn's
HH.MMencoding (10.15is 10:15,15.3is 15:30). The server renders them as readableHH:MMranges. - Day codes are
M,T,W,R,F, whereRis Thursday. - Back-to-back meetings, where one ends exactly as the next begins, are not counted as conflicts.
- This is an unofficial tool that consumes the public Penn Courses API. It caches responses and caps concurrency; please use it respectfully.
License
MIT
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.
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.
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.
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.
E2B
Using MCP to run code via e2b.
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.