Part of the Education & Learning category.
Education and LMS MCP servers are connecting AI agents to learning management systems, study tools, and educational content platforms. Instead of navigating complex LMS APIs or manually managing coursework, these servers let agents access courses, grade assignments, create flashcards, search educational content, and tutor students — all through the Model Context Protocol.
The landscape spans seven areas: LMS platforms (Canvas dominates with 6+ implementations, plus Moodle, Google Classroom, and Brightspace), educational content (e-learning platforms and open educational resource aggregators), spaced repetition (Anki integration), coding education (LeetCode problem access and guided learning), math and tutoring (calculation, visualization, and interactive teaching), academic research (unified scholarly search), and language learning (dictionary tools).
The headline findings: Canvas LMS has more MCP implementations than any other education platform — six independent servers from both instructor and student perspectives. vishalsachdev/canvas-mcp leads with up to 101 tools and 8 agent skills (203 stars — also the most-starred Canvas server) including FERPA-compliant data anonymization and rubric-based bulk grading. DMontgomery40/mcp-canvas-lms provides 50+ tools with a clean TypeScript implementation (103 stars). ankimcp/anki-mcp-server provides the deepest study tool integration with 48 tools (454 stars) for full Anki deck and card management. jinzcdev/leetcode-mcp-server supports both leetcode.com and leetcode.cn (140 stars) with 18 tools including code execution and submission. Instructure launched IgniteAI Agent in March 2026 — the first major LMS vendor to officially adopt MCP as its agent integration standard. No Blackboard, Khan Academy, Coursera, edX, or Duolingo MCP servers exist despite their massive user bases.
LMS Platforms — Canvas
vishalsachdev/canvas-mcp (Most Feature-Rich)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| vishalsachdev/canvas-mcp | 203 | Python | MIT | 101 |
The most comprehensive Canvas MCP server by a wide margin. Provides up to 101 tools and 8 agent skills covering the full Canvas LMS workflow:
- Course management — list courses, access syllabi, modules, pages
- Discussion and announcement tools — create and manage forum interactions
- Assignment tools with analytics — submission tracking, deadline management
- Rubric tools with bulk grading — grade multiple students against rubric criteria in one operation
- FERPA-compliant data anonymization — student data privacy built into the server design
- Code execution environment — run code against the Canvas API directly
Ships with separate Student Guide and Educator Guide documentation. Compatible with Claude Desktop, Cursor, Codex, Windsurf, and 40+ other agents. v1.2.0 (April 2026) added role-based tool filtering via CANVAS_ROLE environment variable, expanded accessibility scanner from 4 to 20 WCAG checks with automated remediation, and Windows compatibility improvements. As of the latest release, v1.11.0 (August 21, 2026), the tool count has grown to “up to 101” and recent releases have focused on protocol-correctness and security hardening rather than new features. Correction (2026-08-26): the previously-noted hosted server at mcp.illinihunt.org has been retired — the project’s README now states a public endpoint without access controls can’t safely expose the server’s built-in code-execution tool, so institutions must install locally or self-host behind their own authentication.
DMontgomery40/mcp-canvas-lms (TypeScript Implementation)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| DMontgomery40/mcp-canvas-lms | 103 | TypeScript | MIT | 50+ |
A TypeScript Canvas MCP server with 50+ tools (Core Student: 33, Instructor: 13, Account Management: 7) organized around core LMS operations:
- Course management — courses, modules, content pages
- Assignments and submissions — create, manage, grade
- Enrollment management — student and instructor role handling
- Grade management — gradebook access and updates
TypeScript implementation with clean API design. Now supports streamable HTTP transport for containerized deployments (Docker/Kubernetes). The MIT license and active maintenance make this the most accessible Canvas server for developers building integrations.
aryankeluskar/canvas-mcp (Canvas + Gradescope)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| aryankeluskar/canvas-mcp | 11 | TypeScript | ISC | 8+ |
Unique dual integration combining Canvas LMS and Gradescope in one server. Natural language queries for courses and assignments, submission status tracking across both platforms. Uses Gemini API for intelligent resource search across course materials.
54yyyu/school-mcp (Canvas + Gradescope + Calendar)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| 54yyyu/school-mcp | 7 | Python | MIT | Multiple |
A student-focused server combining Canvas, Gradescope, and macOS Calendar/Reminders integration. Fetches assignment deadlines from both platforms and syncs them to the system calendar via AppleScript. File download from Canvas included. macOS-only for calendar features.
r-huijts/canvas-mcp (Privacy-First)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| r-huijts/canvas-mcp | 12 | TypeScript | — | 60 |
Available as a Claude Desktop Extension for one-click installation. Privacy-by-default design — student data anonymized automatically. API tokens stored in the OS keychain rather than config files. Covers courses, announcements, rubrics, assignments, modules, pages, and student data.
Other Canvas Servers
- ahnopologetic/canvas-lms-mcp (Python) — lightweight with 22 tools for syllabus, modules, assignments, and grade access
- lucanardinocchi/canvas-mcp (TypeScript) — natural conversation interface for coursework across 6 modules (courses, assignments, submissions, discussions, modules, search)
Instructure IgniteAI Agent (Official — Launched March 2026)
Instructure, the company behind Canvas LMS, announced IgniteAgent at InstructureCon 2025 — a conversational AI assistant that uses MCP as its third-party integration standard. Update (2026-08-26): it has since shipped, now branded IgniteAI Agent, and launched March 12, 2026, initially in the US and Latin America with global availability following — confirmed independently via Instructure’s own newsroom and a PR Newswire distribution of the same release. This makes Instructure the first major LMS vendor to officially adopt the Model Context Protocol; Instructure names Anthropic and OpenAI among the partners building on it. IgniteAI Agent handles multi-step tasks (extending deadlines, converting PDFs, managing course workflows) and lets third-party agents integrate via MCP. It is currently free — no cost for US Canvas customers through June 30, 2026, and free globally through September 30, 2026 to cover the phased international rollout — ahead of a future paid pricing model. This is a significant signal for the future of MCP in education, and no longer just a roadmap promise.
LMS Platforms — Moodle
peancor/moodle-mcp-server (Read + Write)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| peancor/moodle-mcp-server | 42 | JavaScript | MIT | 10 |
The most complete Moodle MCP server, and notably one of the few education servers with full read AND write access:
- list_students — retrieve enrolled students
- get_assignments / get_student_submissions — access coursework
- provide_assignment_feedback — write feedback directly to Moodle
- get_quizzes / get_student_quiz_attempts — quiz data access
- provide_quiz_feedback — write quiz feedback
This write capability is significant — an AI agent can review student submissions and provide feedback directly through the LMS, not just read data. Requires a Moodle API token.
loyaniu/moodle-mcp (Lightweight)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| loyaniu/moodle-mcp | 30 | Python | — | Limited |
A lightweight Moodle server focused on event retrieval via FastMCP. Primarily provides get_upcoming_events for deadline tracking. Early development stage with limited functionality.
LMS Platforms — Google Classroom & Brightspace
faizan45640/google-classroom-mcp-server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| faizan45640/google-classroom-mcp-server | 4 | JavaScript | MIT | 3 |
Basic Google Classroom integration with three tools: course listing, course details, and assignment access. Requires Google API credentials and OAuth setup. Given Google Classroom’s massive deployment in K-12 education, the minimal MCP coverage here is a notable gap.
pranav-vijayananth/brightspace-mcp-server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| pranav-vijayananth/brightspace-mcp-server | 7 | Python | Apache-2.0 | Limited |
A creative solution for D2L Brightspace — since students lack official API access, this server uses Playwright web scraping to extract course lists, assignments, and grades. Includes automated Duo Mobile 2FA support. Built by a Purdue University student specifically for their institution’s Brightspace instance. Fragile by nature (web scraping), but demonstrates real student demand for LMS AI integration.
Educational Content & Courseware
EduBase/MCP (Official E-Learning Platform)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| EduBase/MCP | 27 | TypeScript | MIT | Many |
Official MCP server from the EduBase e-learning platform, exposing the full platform API as MCP tools (each endpoint becomes a tool named edubase_<method>_<endpoint>). Key capabilities:
- Advanced quiz system — parametric question generation, randomization, auto-grading
- Cheating detection — real-time monitoring during assessments
- SCORM module support — industry-standard courseware compatibility
- SSO integration — institutional authentication
- GDPR compliance — data protection built in
- Triple transport — stdio, SSE, and streamable HTTP
Supports Docker deployment for self-hosted instances. Very actively maintained, with releases running through v1.2.9 (August 2026). Also installable via Smithery or remote MCP server.
Cicatriiz/openedu-mcp (Open Educational Resources)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| Cicatriiz/openedu-mcp | 10 | Python | — | 21+ |
An educational resource discovery server aggregating four free APIs into a unified interface:
- OpenLibrary — book search and metadata
- Wikipedia — encyclopedia article access
- arXiv — academic paper search
- Dictionary API — word definitions and usage
Includes grade-level appropriateness filtering for K-12 contexts and curriculum planning support. Available via stdio, HTTP, and SSE transports.
odewahn/orm-discovery-mcp (Built by O’Reilly’s CTO)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| odewahn/orm-discovery-mcp | 4 | Python | — | 1 |
Built by O’Reilly CTO Andrew Odewahn, this server provides search access to O’Reilly’s extensive technical book and article library via a single search_content tool. Returns structured metadata including titles, authors, ratings, and direct URLs. Released March 2025. Requires an O’Reilly subscription for full content access.
EduChain MCP Servers (AI Content Generation)
Multiple community implementations wrapping the educhain Python library for AI-powered educational content generation:
- MCQ Generator — create multiple-choice questions with options, correct answers, and explanations
- Lesson Plan Generator — structured lesson plans from topics
- Flashcard Generator — study material creation
Uses local LLMs (Mistral-7B-Instruct) for generation. Implementations include abanindra3/educhain-mcp and yogasairam/educhain-mcp-server-final. Quality depends on the underlying model.
Education Data
ckz/edu_data_mcp_server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| ckz/edu_data_mcp_server | 2 | Python | — | 2 |
Accesses the Urban Institute’s Education Data API for US education statistics:
- get_education_data — query enrollment, school/district directories, financial data
- get_education_data_summary — aggregation and summary statistics
Covers schools, districts, colleges, and universities using data from CCD, IPEDS, and CRDC datasets. Read-only, US education data only.
Spaced Repetition & Study Tools
ankimcp/anki-mcp-server (Most Complete Study Tool)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| ankimcp/anki-mcp-server | 454 | TypeScript | MIT | 48 |
The most feature-rich study tool MCP server with 48 tools for full Anki integration:
- Deck management — sync, create, list, organize decks
- Card review — study cards through the MCP interface
- Note and card management — create, modify, delete with media file support
- Evidence-based creation — includes a
twenty_rulesprompt implementing SuperMemo’s 20 Rules of Formulating Knowledge for optimal flashcard design - Dual modes — STDIO for desktop clients, Web mode (with integrated ngrok tunneling) for Claude.ai and ChatGPT
Earlier releases added bulk note creation (up to 100 notes per call) and MCP Tool Annotations with structured metadata; v0.22.5 replaced a blocklist-based media-URL guard with a fail-closed allowlist to close an SSRF vulnerability. Most recently, v0.24.0 (August 13, 2026) upgraded to MCP SDK v2, and v0.24.1 (August 16, 2026, current latest) fixed tunnel-mode idle-timeout handling — the tool count has held steady at 48 through all of these. Published on npm as @ankimcp/anki-mcp-server. Requires Anki desktop running locally.
Other Anki Servers
- nailuoGG/anki-mcp-server (Python) — AnkiConnect-based with deck management and note creation
- scorzeth/anki-mcp-server — card review and creation via AnkiConnect
- johwiebe/anki-mcp (Python) — basic Anki MCP integration
The presence of 4+ Anki MCP servers reflects strong demand from the spaced repetition community for AI-assisted flashcard creation and study workflows.
Coding Education
jinzcdev/leetcode-mcp-server (Most Popular)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| jinzcdev/leetcode-mcp-server | 140 | TypeScript | MIT | 18 |
The most starred coding education MCP server, supporting both leetcode.com (Global) and leetcode.cn (China) with 18 tools across five categories:
- Problems (3 tools) — question details, constraints, examples, advanced search by tags/difficulty/topic
- Users (7 tools) — profiles, submission history, solved problems (with optional authentication)
- Submissions (2 tools) — submission retrieval and history
- Notes (4 tools) — personal notes management
- Solutions (2 tools) — community and official solutions
v1.3.0 added run_code and submit_solution tools — agents can now execute code against LeetCode test cases and submit solutions directly, not just read problems. The current latest release, v1.4.0 (July 2026), upgraded the MCP SDK and added Streamable HTTP transport; the tool count remains 18. Published on NPM for easy installation. The dual-region support makes this the only coding education server serving both English and Chinese-speaking developer communities.
SPerekrestova/interactive-leetcode-mcp (Pedagogical)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| SPerekrestova/interactive-leetcode-mcp | 11 | TypeScript | MIT | 18 |
Takes a deliberately pedagogical approach to coding education. Instead of just providing answers, it enforces progressive learning:
- 4-level hint system — gives increasingly specific hints before revealing solutions
- Pedagogical best practices — encourages independent problem-solving
- Agent skill — structured session flow for guided DSA (Data Structures & Algorithms) learning
Now at v3.2.5 with multi-language support (Java, Python, C++, JavaScript, TypeScript) and Claude Code/ClawHub integrations. This design philosophy — making AI assist learning rather than bypass it — is a standout in the education MCP space.
doggybee/mcp-server-leetcode
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| doggybee/mcp-server-leetcode | 44 | TypeScript | — | Multiple |
LeetCode access via GraphQL with metadata filtering, user profiles, submission history, and contest data. Another option for developers who want LeetCode integration without the pedagogical constraints of interactive-leetcode-mcp.
Math & Tutoring
clouatre-labs/math-mcp-learning-server
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| clouatre-labs/math-mcp-learning-server | 5 | Python | Apache 2.0 | 17 |
A dedicated math education server with 17 tools across workspace, math, matrix, and visualization categories:
- calculate / statistics / compound_interest / matrix_multiply — math operations
- convert_units — unit conversion
- plot_function / create_histogram / plot_line_chart / plot_scatter_chart / plot_box_plot / plot_financial_line — data visualization
- math_tutor — interactive math tutoring prompts
- formula_explainer — break down mathematical formulas step by step
Built with FastMCP 2.0 and the official MCP Python SDK. Includes persistent workspace for ongoing calculations. Available on PyPI and the MCP Registry, with cloud hosting at math-mcp.fastmcp.app/mcp. Holds OpenSSF Best Practices Silver certification — unusually high quality bar for a project this small. Now at v0.12.3 (August 2026).
moarshy/mcp-tutor (Documentation-to-Lessons)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| moarshy/mcp-tutor | 0 | Python | — | 4 tools, 6 prompts |
Transforms GitHub documentation repositories into interactive educational content. Uses document-level processing with tool-augmented generation (TAG) for context-aware teaching. An interesting approach — turning existing open-source documentation into structured lessons without manual content creation.
Academic Research Tools
aringadre76/mcp-for-research (Unified Search)
| Server | Stars | Language | License | Tools |
|---|---|---|---|---|
| aringadre76/mcp-for-research | 14 | JavaScript | MIT | 5 |
Consolidates PubMed, Google Scholar, arXiv, and JSTOR into a unified search interface. Replaced 24 individual endpoint tools with 5 consolidated tools for a cleaner agent experience. Published on NPM.
Other Academic Servers
- JackKuo666/Google-Scholar-MCP-Server (Python) — Google Scholar search with paper access
- zongmin-yu/semantic-scholar-fastmcp-mcp-server (Python) — Semantic Scholar API for paper data, author info, and citation network exploration
- afrise/academic-search-mcp-server — dual Semantic Scholar and Crossref integration
- MicrosoftDocs/mcp — official Microsoft Learn documentation server with search, article fetch, and code samples (free, no auth required)
Language Learning
meowrain/mcp-server-cambridge-dict
Cambridge Dictionary integration providing word lookup with pronunciation, definitions, and example sentences.
Traves-Theberge/Word_of_the_day
Daily vocabulary learning via the free Dictionary API with word definitions and random vocabulary features.
What’s Missing
The gaps in education MCP servers are striking:
- Blackboard/Ultra — no MCP server despite being the second-largest LMS globally
- Khan Academy — no MCP server despite being the most popular free learning platform
- Coursera / edX — no MCP servers for the two largest MOOC platforms
- Duolingo / Memrise — no language learning app integration
- PowerSchool — no student information system MCP servers
- Turnitin — no plagiarism detection integration
- Classroom management (attendance, behavior tracking) — completely absent
The concentration around Canvas (6+ servers) versus the absence of Blackboard (0 servers) likely reflects Canvas’s developer-friendly API and the university CS student demographic most likely to build MCP servers.
What’s New (May 2026)
Canvas MCP keeps growing. vishalsachdev/canvas-mcp has expanded since our last refresh — from 92 tools at v1.2.0 (April 2026) to up to 101 tools and 8 agent skills as of v1.11.0 (August 21, 2026, its current latest release), with recent releases (v1.6.0 through v1.11.0, July–August 2026) focused mainly on protocol-correctness and security fixes. The server now works with Claude Desktop, Cursor, Codex, Windsurf, and 40+ other agents. Canvas MCP is compliant with Canvas LMS API 2024–2026 requirements (User-Agent header, per_page pagination).
Docebo MCP Server: beta to GA. Docebo AgentHub launched on April 21, 2026, with the Docebo MCP Server entering public beta. It makes Docebo a native knowledge source inside Claude, Microsoft Copilot, ChatGPT, and any other MCP-enabled AI tool — agents can look up learner progress, surface course recommendations, and check certification status without custom integration. Update (2026-08-26): the server reached general availability in July 2026 as planned, adding admin tools for managing enrollments, building courses, and curating Go1 content directly from Claude, ChatGPT, Copilot, and Gemini. Docebo is an enterprise LMS used primarily in corporate training contexts, adding a fifth major platform to the education MCP ecosystem after Canvas, Moodle, Brightspace, and Google Classroom.
Canvas/Instructure breach (May 2026). The hacking group ShinyHunters targeted Instructure — the company behind Canvas LMS — claiming responsibility on May 3 and launching a public extortion campaign with a May 7 deadline; Instructure reportedly paid a ransom around May 11, 2026, after confirming exposure of names, email addresses, student IDs, and some private messages. This is relevant for any institution using Canvas MCP servers that authenticate via Canvas API tokens: if API tokens were exposed in the breach, they should be rotated immediately. This incident underscores the importance of scoped, read-only API keys when deploying Canvas MCP servers.
Agentic grading reaches LMS integrations. AI grading tools are now operating inside Canvas, Moodle, and Blackboard via LTI grade passback — AI evaluates submissions with institutional reference examples, and grades flow back automatically. This is the LMS side of what the rubric_assessment tools in Canvas MCP servers enable: not just surfacing rubrics, but closing the loop with automated grade submission.
The Bottom Line
Rating: 3.5 / 5 — Canvas LMS integration is genuinely impressive, with the deepest MCP coverage of any single education platform. Spaced repetition (Anki) and coding education (LeetCode) are well-served. The pedagogical approach of interactive-leetcode-mcp — progressive hints rather than instant answers — shows thoughtful design for learning contexts. EduBase’s official server brings enterprise e-learning features. However, the category suffers from massive platform gaps: the biggest names in education (Blackboard, Khan Academy, Coursera, Duolingo) have zero MCP presence. Google Classroom, used by hundreds of millions of students globally, has just 3 tools. Most servers are student projects rather than vendor-supported products. The bright spot: Instructure’s IgniteAgent announcement signals that LMS vendors are beginning to take MCP seriously, which could catalyze broader adoption in 2026.
ChatForest independently researches MCP servers using publicly available information — GitHub repositories, documentation, and community discussions. We do not have hands-on access to every server reviewed. Star counts reflect the time of research and may have changed. Always verify server status and compatibility before adopting any MCP server for your workflow.
This review was last updated on 2026-05-15 using Claude Sonnet 4.6 (Anthropic).