ntnu-mcp
A read-only MCP server exposing NTNU course data to LLM assistants, enabling course catalog search, teaching schedules, weekly timetables, grade statistics, and course comparison tools.
README
ntnu-mcp
A remote MCP server that exposes NTNU (Norwegian University of Science and Technology) course data to LLM assistants; course catalog search, teaching schedules, weekly timetables, grade statistics, course descriptions, exam logistics, and study programs with their official study plans, plus comparison tools for students choosing between courses. Runs as a Cloudflare Worker; no installation required on the client side.
Tools
Twelve tools, all read-only:
search_courses— free-text search of NTNU's course catalog for a given year; returns codes, names, campus, and exam dates.get_course_schedule— every dated teaching activity (lectures, labs) for a course in a year, with Oslo-local times and rooms.get_weekly_timetable— the recurring weekly timetable grid for a course: weekday, times, ISO-week ranges, and rooms.get_grade_distribution— historical grade distribution for a course, per year and semester, with counts and percentages.get_course_versions— the DBH-versioned course codes (e.g.TDT4100-1) behind a bare course code, useful when grade lookups come back empty.get_semesters— NTNU's terms with ids like26h/26v, teaching weeks, exam-period dates, and which term is current.get_course_info— everything about a course except exam logistics: credits, level, campus, prerequisites, mandatory activities, content/learning outcomes, credit reductions, the study programs the teaching is planned for, contacts, and alert notices. English by default,language: "nb"for Norwegian.get_exam_info— exam logistics for up to 10 courses in one call: every occasion (ordinary/re-sit) with date, start time, duration, permitted-aids code ("hjelpemiddelkode") with its meaning, exam system, and assigned rooms.search_study_programs— search NTNU's ~400 study programs by text, level, and city; returns program codes forget_study_plan.get_study_plan— the official plan for a program and cohort intake year: courses (codes, credits, mandatory/elective) per semester, with specialization choice points — the bridge from "I study X" to concrete course codes.compare_courses— side-by-side comparison of several courses: name, campus, exam dates, weekly teaching hours, and latest grade distribution.check_timetable_conflicts— pairwise check for weekly-schedule clashes and exam-date collisions across a set of courses, before registering for them.
Connecting
Paste the server's MCP URL into Claude's or ChatGPT's custom connector settings:
https://ntnu-mcp.martinsundal.no/mcp
The server requires no authentication, since it only serves public, read-only NTNU course data.
The server is also listed in the
official MCP registry as
io.github.MartinSA04/ntnu-mcp (see server.json); clients
that browse the registry can discover it there. It is an unofficial
community server, not affiliated with NTNU. Registry updates are
tag-driven: npm version <patch|minor|major> && git push --follow-tags —
the version script keeps server.json in sync, and the
release workflow runs the check suite,
deploys the Worker, re-publishes the registry entry (GitHub OIDC), and
creates the GitHub Release.
Architecture
All NTNU upstream knowledge — endpoints, retry policy, parsing of the
Liferay catalog, DBH grade statistics, and TP semester data — lives in the
ntnu-api TypeScript client
library, which this repo depends on as an npm package. Upstream fixes and
new data sources land in ntnu-api, not here. This repo's only job is to
call that typed client and shape its output for LLM consumption: English
names with Norwegian fallback, Oslo-local timestamps, grade rows collapsed
into per-term distribution tables, and guidance notes on empty results. This
layering is set out in ntnu-api's
TypeScript migration spec.
To keep load on NTNU's servers minimal, every upstream call is cached in two tiers (per-isolate memory in front of a shared Workers KV namespace), so each resource is fetched roughly once per TTL globally: catalog searches, timetables, and schedules for 1 hour; grade statistics and the semester list for 24 hours. KV failures degrade to memory-only caching, never to tool errors. In tests the cache runs memory-only.
Local development
Requires Node 22 (pinned via mise).
mise install # installs the pinned Node version
npm install
Common tasks, available both as mise tasks and npm scripts:
mise run dev # npm run dev — wrangler dev, local worker with hot reload
mise run test # npm test — vitest, over captured fixtures (no live network)
mise run typecheck # npm run typecheck — tsc --noEmit
mise run lint # npm run lint — biome check
mise run fmt # npm run fmt — biome check --write
mise run check # lint + typecheck + test
Tests run under plain Node (not the Workers pool) with vitest, driving the
tool functions directly against fetch fixtures captured verbatim from
ntnu-api's test suite — no calls to real NTNU services are made in CI or
locally.
Deploy
Pushes to main deploy automatically (the repo is connected to Cloudflare
Workers Builds). For a manual deploy from a checkout:
npm run deploy # wrangler deploy
Runs on Cloudflare's free plan: the MCP session state is backed by a Durable Object using SQLite storage, which is available on the free tier (no Workers Paid subscription required unless per-request CPU limits are hit at scale).
License
MIT — see LICENSE.
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.