Part of the Education & Learning category.
Education MCP servers are connecting AI agents to learning management systems, flashcard applications, academic research databases, assessment platforms, and STEM computation tools. Instead of manually navigating LMS dashboards, searching academic databases, or creating flashcards one at a time, these servers let AI assistants manage coursework, track grades, search research papers, run spaced repetition sessions, generate quizzes, and solve mathematical problems — all through the Model Context Protocol.
The landscape spans seven areas: LMS integration (Canvas, Moodle, Google Classroom, D2L Brightspace), flashcards & spaced repetition (Anki implementations), academic research (arXiv, Semantic Scholar, multi-platform aggregators), enterprise learning (Udemy Business), assessment & tutoring (EduBase, quiz generators), STEM tools (Wolfram Alpha computation), and knowledge reference (Wikipedia).
The headline findings: vishalsachdev/canvas-mcp has overtaken DMontgomery40 as the most-starred Canvas server — hitting 107 stars (+43%) with v1.2.0’s accessibility scanner expansion. Academic research is the fastest-growing subcategory — paper-search-mcp surged to 1,200 stars (+52%) and now aggregates 20+ sources, while arXiv MCP hit 2,600 stars with new semantic search tools. Anki MCP grew 32% to 237 stars — the strongest flashcard integration in any MCP category. Udemy Business MCP went GA on March 31, 2026 — the first publicly-traded EdTech company shipping production MCP infrastructure. D2L Brightspace now has 3+ MCP implementations — up from a single server. Khan Academy gets its first MCP server — no longer absent from the ecosystem. 52 education servers on PulseMCP. The biggest remaining gap: Blackboard still has zero MCP implementations.
Ecosystem stats (April 2026)
| Metric | Value |
|---|---|
| Education servers on PulseMCP | 52 |
| LMS platforms with MCP servers | 4 (Canvas, Moodle, D2L, Google Classroom) |
| Highest-starred server | arxiv-mcp-server (2,600) |
| Fastest-growing server | paper-search-mcp (+52% in 38 days) |
| Official vendor servers | 1 (Udemy Business, GA) |
| AI-in-education bills (2026) | 134 across 31 US states |
LMS Integration
Canvas LMS — DMontgomery40/mcp-canvas-lms
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| DMontgomery40/mcp-canvas-lms | ~96 | TypeScript | — | 50+ tools |
Comprehensive Canvas coverage across three user types:
- Students — course access, assignment submission, grade tracking, discussion participation, file management, calendar events
- Instructors — course creation, assignment management, grading, student enrollment, quiz administration, rubric handling
- Account administrators — user management, account oversight, reporting, sub-account administration
Version 2.3.0 introduced streamable HTTP transport support, Docker deployment options, and automatic retries with pagination. The server supports both stdio and HTTP transport modes.
Canvas LMS — vishalsachdev/canvas-mcp
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| vishalsachdev/canvas-mcp | ~107 | Python/TypeScript | MIT | 88 tools + 8 agent skills |
Now the most-starred Canvas implementation (+43% growth), with FERPA-compliant data anonymization for educator use — a critical differentiator for institutional deployment. v1.2.0 (April 2026) added role-based tool filtering, expanded the accessibility scanner from 4 to 20 checks, security hardening, and Windows compatibility improvements.
Offers 8 agent skills via skills.sh for 40+ coding agents, a code execution API for token-efficient bulk operations, and a hosted server option requiring no local installation. Supports Claude Desktop, Cursor, Zed, Windsurf, Continue, Replit, Copilot Studio, and other MCP-compatible clients. The learning designer capabilities include course quality control and accessibility auditing — features no other Canvas MCP server offers.
Other Canvas Implementations
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| r-huijts/canvas-mcp | ~12 | TypeScript | — | 25+ tools |
| ahnopologetic/canvas-lms-mcp | — | TypeScript | — | 22 tools |
| akshsgaur/CMUCanvasMCPServer | — | — | — | CMU-specific |
| lucanardinocchi/canvas-mcp | — | TypeScript | — | Multiple |
| mtgibbs/canvas-lms-mcp | — | TypeScript | — | Multiple |
At least five additional Canvas MCP servers exist, ranging from minimal read-only grade checkers to university-specific implementations (CMU). r-huijts/canvas-mcp grew to 12 stars with privacy-first anonymization and a PulseMCP listing. The fragmentation reflects Canvas’s dominance in higher education — Instructure powers over 6,000 institutions, making it the most targeted LMS for MCP integration.
Moodle MCP Server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| peancor/moodle-mcp-server | ~36 | JavaScript | MIT | 9 tools |
The leading Moodle MCP server provides tools across four management categories:
- Course management —
get_courseswith pagination - Student oversight —
list_studentswith ID, name, email, last access - Assignment handling —
get_assignments,get_student_submissions,provide_assignment_feedback - Assessment —
get_quizzes,get_quiz_attempts,provide_quiz_feedback
Requires Node.js v14+ and a Moodle API token with appropriate permissions. A second implementation, loyaniu/moodle-mcp (16 stars), offers a Python alternative focused on upcoming events.
The grading and feedback tools make this genuinely useful for instructors managing large courses — an AI assistant can review submissions, generate feedback drafts, and submit grades through natural language commands.
Google Classroom MCP Server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| faizan45640/google-classroom-mcp-server | ~3 | JavaScript | MIT | 3 tools |
A minimal implementation providing read-only access to Google Classroom data:
- courses — list all courses
- course-details — detailed course info with announcements
- assignments — assignment data for a course
Uses OAuth 2.0 authentication with automatic token refresh. Given Google Classroom’s massive adoption in K-12 education (150M+ users), the limited tool count here represents a significant gap — there’s no assignment submission, grade posting, or student roster management.
D2L Brightspace MCP Servers
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| RohanMuppa/brightspace-mcp-server | ~14 | — | MIT | 7+ areas |
| joshuasoup/d2l-mcp | — | — | — | 12 tools |
| bencered/d2l-mcp-server | — | — | — | 12 tools |
D2L Brightspace went from a single MCP implementation to three competing servers — the fastest LMS expansion in this review cycle. RohanMuppa’s server (14 stars, created Feb 2026) is the most-starred, with v1.0.4, AES-256-GCM credential encryption, and compatibility with any school running D2L Brightspace. joshuasoup/d2l-mcp adds file downloads and text extraction. bencered’s original provides automated SSO authentication and persistent session storage.
A student-focused implementation by pranav-vijayananth/brightspace-mcp-server at Purdue University also exists. D2L Brightspace is also available on the Pipedream MCP marketplace.
The ethical note in bencered’s repository is worth highlighting: “This should NOT be used to cheat on any assignments or to gain an unfair advantage… only intended to enable agents to better understand assignment scheduling and to help with learning of course material.”
Flashcards & Spaced Repetition
Anki MCP Server — ankimcp
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| ankimcp/anki-mcp-server | ~237 | TypeScript | AGPL-3.0 | 21+ tools |
The most comprehensive Anki integration, with tools organized into five categories:
- Review & Study (4 tools) — quiz sessions, card scheduling, review tracking
- Deck Management (4 tools) — create, organize, and configure decks
- Note Management (6 tools) — CRUD operations with batch support (up to 100 notes), Anki query syntax
- Media Management (4 tools) — audio/image handling with multiple upload methods
- Model/Template Management (3 tools) — template configuration with HTML/CSS awareness
Deployment flexibility is a standout feature: local stdio, HTTP server, or ngrok-tunneled remote access for web-based AI tools. Includes a read-only safety mode to prevent accidental modifications. Still in beta with APIs subject to change.
Anki MCP — arielbk
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| arielbk/anki-mcp | ~9 | TypeScript | MIT | 7 tools |
A lighter Anki implementation focused on conversational flashcard management. Features include quizzing on specific cards, creating flashcards about topics, turning PDFs into spaced repetition cards, generating cloze deletion cards from lecture notes, deep analytics on learning progress, and bulk operations for thousands of cards.
Other Anki Implementations
| Server | Language | License | Focus |
|---|---|---|---|
| dhkim0124/anki-mcp-server | Python | MIT | Natural language card creation |
| nailuoGG/anki-mcp-server | TypeScript | — | AnkiConnect bridge |
| ezynda3/anki-mcp | — | MIT | Programmatic card management |
The Anki ecosystem mirrors Canvas — multiple implementations competing on different axes. The ankimcp version wins on features and stars; arielbk/anki-mcp wins on simplicity. All require the AnkiConnect plugin running locally.
Academic Research
arXiv MCP Server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| blazickjp/arxiv-mcp-server | ~2,600 | Python | MIT | 8 tools |
The highest-starred education MCP server by a wide margin, now at v0.4.11 (April 3, 2026) with 8 tools — double the original four. The core tools remain (paper search, download, list, read), but four experimental tools have been added:
- Semantic Search — vector-based similarity search across downloaded papers
- Citation Graph — explore citation relationships between papers
- Watch Topic — set up alerts for new papers in specific areas
- Check Alerts — review papers matching watched topics
Includes a set of research prompts for comprehensive paper analysis workflows. The star count reflects broad adoption beyond education — researchers, developers, and AI practitioners all use arXiv daily. Note: the license is MIT (previously listed as Apache-2.0 — corrected per LICENSE file and PyPI).
Paper Search MCP
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| openags/paper-search-mcp | ~1,200 | Python | MIT | 20+ sources |
The fastest-growing education MCP server — stars surged from 791 to 1,200 (+52%) in 38 days. Now aggregates 20+ academic sources (up from 7), including:
- arXiv, PubMed, bioRxiv, medRxiv, Google Scholar, IACR ePrint Archive, Semantic Scholar, Crossref, OpenAlex, and more
Uses a free-first strategy with OA-first fallback chain, standardized Paper class for consistent output across platforms, async request handling via httpx, and an extensible architecture. Optional Sci-Hub integration for full paper downloads. Claude Code skills integration added.
Academic Search MCP Server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| afrise/academic-search-mcp-server | — | Python | — | Multiple |
Focused on Semantic Scholar and Crossref integration, providing comprehensive metadata including title, authors, year, DOI, venue, open access status, PDF URLs, abstracts, and TL;DR summaries. Citation generation in BibTeX and APA formats.
Enterprise Learning
Udemy Business MCP Server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| Udemy Business MCP | — | — | Commercial | Multiple |
The first publicly-traded EdTech company to ship a production MCP server. Udemy Business AI Connectors went GA on March 31, 2026 — moving from preview to general availability. Implements MCP spec version 2025-11-25 with OAuth and static client credentials. Key capabilities:
- On-demand support — real-time problem-solving via micro-courses and curated lessons
- Plug-and-play integration — pre-built MCP connectors for ChatGPT, Claude, Cursor, custom enterprise agents
- Role-specific contextualization — content matched to user’s role and learning context
- Intelligent content matching — relevant course recommendations within existing AI tools
This is an enterprise product aimed at corporate L&D teams (paid add-on, pricing undisclosed), not individual learners. The GA launch is the most significant commercial EdTech MCP deployment — a publicly-traded company with full production MCP infrastructure validates the protocol’s relevance beyond developer tooling.
Assessment & Tutoring
EduBase MCP Server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| EduBase/MCP | ~25 | TypeScript | MIT | Dynamic (API-based) |
Connects AI to the EduBase e-learning platform with features that matter for institutional deployment:
- Advanced quiz system with parametrization and LaTeX support
- Unified learning environment: videos, exams, documents, SCORM modules
- Enterprise-grade security with SSO and GDPR compliance
- Multiple transport protocols (stdio, SSE, streamable HTTP)
- AI-assisted content transformation — turn existing materials into interactive assessments
Each documented API endpoint is exposed as a separate tool (edubase_<method>_<endpoint>), making the tool count dynamic based on the platform version.
Assessment Suite MCP
An MCP server enabling AI-generated quiz export in multiple formats (CSV, PDF, JSON, Markdown). Bridges the gap between AI question generation and classroom-ready assessment materials — educators can generate questions conversationally and export them directly to their preferred format without manual copy-pasting.
STEM Computation
Wolfram Alpha MCP Servers
| Server | Language | Focus |
|---|---|---|
| Garoth/wolframalpha-llm-mcp | TypeScript | Structured knowledge + math |
| StoneDot/wolframalpha-mcp-server | TypeScript | Computational queries |
| akalaric/mcp-wolframalpha | Python | LangChain integration |
| cnosuke/mcp-wolfram-alpha | Go | API access |
Multiple implementations exist for connecting AI to Wolfram Alpha’s computational engine. The core value proposition is delegating numerical computation that LLMs struggle with — derivatives, integrals, equation solving, statistics, physics constants, unit conversions, chemical reactions. All require a Wolfram Alpha API key.
For STEM education specifically, these servers enable AI tutors to show their work: an LLM can explain the concepts while Wolfram Alpha handles the actual computation, combining pedagogical reasoning with mathematical precision.
Wikipedia MCP Servers
Several implementations provide AI access to Wikipedia’s knowledge base, with features including article search, content retrieval, section extraction, related topic discovery, and multi-language support. Useful as a reference layer for educational AI applications — grounding explanations in encyclopedic sources rather than relying solely on the LLM’s training data.
New Entrants (Since March 2026)
OpenEdu MCP
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| Cicatriiz/openedu-mcp | ~7 | — | — | 21+ tools |
A curriculum-aligned educational MCP server pulling from Open Library, Wikipedia, Dictionary API, and arXiv. The key differentiator: grade-level filtering from K-2 through College, with curriculum alignment to Common Core and NGSS (Next Generation Science Standards). Includes SQLite caching for performance. The first MCP server to explicitly target K-12 curriculum standards — a notable gap-filler.
Khan Academy MCP
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| aicoder2009/khanacademyMCP | ~1 | — | MIT | 12 tools |
The first Khan Academy MCP server — tools include search, list_subjects, get_topic_tree, get_content, get_course, and get_transcript. No API key required. Very early stage (1 star), but fills a conspicuous gap — Khan Academy is one of the largest free educational platforms (120M+ registered users) and was previously absent from the MCP ecosystem entirely.
Additional New Servers on PulseMCP
- Smartschool MCP (by MauroDruwel) — course, grades, and assignment management for the Belgian Smartschool platform
- OpenEMIS MCP (by tixuz) — education management information system integration
- Duolingo Game Status MCP (by Sniffleupagus, March 24, 2026) — streaks, XP, language courses (note: Duolingo’s internal 180+ MCP tools remain enterprise-only)
AI-in-Education Policy Context
A regulatory wave is creating both urgency and uncertainty for education MCP adoption:
- 134 AI-in-education bills introduced across 31 US states in the 2026 legislative session
- Ohio mandates all public school districts adopt formal AI policies by July 1, 2026
- Maryland requires AI guidance, AI literacy integration, AI coordinators, and university-certified compliant AI tools
- 50-84% of K-12 students already use AI for schoolwork; ~90% in higher education
- Only 13% of schools have formal AI policies despite 87% teacher interest
- Only 48% of institutions have formal AI policies; just 6% of teachers say those policies are clear
The policy fragmentation may slow institutional MCP adoption (IT departments hesitate without clear guidance), but mandates like Ohio’s push districts toward structured, auditable AI tooling — exactly what MCP provides.
What’s missing
The education MCP ecosystem has significant gaps:
- Major LMS platforms — Blackboard, Sakai, Open edX, Schoology have zero MCP implementations
- MOOC platforms — no Coursera, LinkedIn Learning, or edX integrations (Udemy GA, Khan Academy minimal at 1 star)
- Student information systems — no Banner, PowerSchool, Infinite Campus, or Ellucian SIS connectivity
- Library systems — no OCLC WorldCat, Ex Libris Alma, or MARC catalog integrations
- Academic integrity — no Turnitin, Grammarly, or plagiarism detection tools
- Adaptive learning — no adaptive engines (Knewton, ALEKS, DreamBox) have MCP support
- Standardized testing — no SAT, GRE, AP, or state assessment integrations
- Learning analytics — no xAPI/Caliper, learning record stores, or analytics dashboards
- Proctoring — no Respondus, ExamSoft, or ProctorU integrations
- K-12 specific — Google Classroom support is minimal (3 read-only tools for a platform with 150M+ users)
The bottom line
The education MCP server ecosystem is fragmented but growing fast. In 38 days: paper-search-mcp surged 52% to 1,200 stars, vishalsachdev/canvas-mcp overtook DMontgomery40 with 43% growth and a major v1.2.0 release, Anki MCP grew 32% to 237 stars, D2L Brightspace went from one to three implementations, and PulseMCP now lists 52 education servers.
The biggest development: Udemy Business MCP went GA on March 31, 2026 — the first publicly-traded EdTech company shipping production MCP infrastructure. This is no longer a “signal” — it’s a shipped product with OAuth, role-specific contextualization, and enterprise agent compatibility. The question is now when (not if) Coursera and LinkedIn Learning follow.
New entrants are filling gaps: OpenEdu brings the first K-12 curriculum-aligned MCP server (Common Core, NGSS), Khan Academy gets its first MCP integration (early stage), and the D2L Brightspace expansion means four of the five major LMS platforms now have MCP support.
The policy landscape is creating both tailwinds and headwinds: 134 AI-in-education bills across 31 states, with Ohio mandating formal AI policies by July 2026. This pushes districts toward structured, auditable AI tooling — but the fragmentation also creates uncertainty.
The ecosystem remains deeply incomplete. The three pillars of institutional education technology — LMS (partially covered), SIS (not covered), and library systems (not covered) — need all three for a complete campus AI workflow. Blackboard’s absence is conspicuous. Academic integrity tools (Turnitin) and adaptive learning engines are still missing entirely.
Rating: 3.5/5 — Academic research tools are excellent (arXiv 2,600 stars, paper-search-mcp 1,200 stars), Canvas competition is healthy, Anki is mature, Udemy GA adds real commercial credibility, and new entrants (OpenEdu, Khan Academy, D2L growth) show the ecosystem broadening. But Blackboard’s absence, missing SIS/library integrations, and thin K-12 coverage (despite OpenEdu’s promising start) prevent a higher score. The growth trajectory is strong — this category could reach 4.0 if Blackboard or a major SIS vendor ships MCP support.
This review was last refreshed on 2026-04-22 using Claude Opus 4.6 (Anthropic).