Best MCP Servers for Education in 2026

The best MCP servers for education in November 2026 are a short list, and that is the honest answer. Google Classroom, Panopto, Wiley, Learning Commons, Open edX, Teachfloor and Microsoft Learn publish their own MCP servers. Canvas, the most-asked-about platform, still relies on a community server for MCP access.

MCP (Model Context Protocol) is the open standard that lets an AI app such as Claude or ChatGPT read from and act in outside software. In a school or training team, that can mean "List every ungraded submission in my Biology course" or "Find the lecture where we covered photosynthesis." Some of these servers only read. Others can post announcements, change grades or enroll learners.

We checked every entry against the vendor's or project's own page on 2 November 2026. eSEOspace did not test these products. For the wider picture of where MCP fits in schools, universities and corporate training, read our guide to MCP for education.

How we chose

We used the MCP Guide's standard listicle criteria:

• Live and public. The vendor or project must publish a page or repository for its MCP server. We opened each one on 2 November 2026.
• Clear pricing. Prices come only from official pricing pages. Where none exists, we say "Pricing on request" or "Free and open source."
• Security basics. The docs must describe sign-in, permissions, read-only behavior or audit logging.

Education has fewer credible MCP servers than most categories, so this list has eight entries rather than twelve. We included one community project (Canvas MCP) because Canvas is so widely used and the project documents its privacy controls. The order groups similar tools; it is not a ranking. Browse other lists on the MCP Guide's best-of hub.

Comparison at a glance

1. Google Classroom MCP: official, but a preview

What it is. Google hosts a Classroom MCP server at https://classroommcp.googleapis.com/mcp/v1. It has 16 tools, including list_courses, get_course_roster, list_course_work_submissions, create_course_work, create_announcement and get_guardian_info. Google says it is "Available as part of the Google Workspace Developer Preview Program."

Best for: Teachers and school IT teams that run classes in Google Classroom.

Pricing: No price is listed for the MCP server. Access requires the Developer Preview Program and a Google Cloud project.

Pros:

• Built and hosted by Google, with OAuth sign-in.
• Google's setup guide covers Claude as a custom connector and Google Antigravity.
• Includes rubrics, topics and course standards, not just assignments.

Cons:

• It can create and update coursework and announcements, so a bad prompt can reach students.
• Setup means a Cloud project, an OAuth client and preview enrollment, which is a lot for a single teacher.

Checked against: Google Classroom MCP reference on 2 November 2026. Related profile: Google Workspace MCP server.

2. Panopto MCP Server: read-only access to lecture video

What it is. Panopto's MCP Server is an add-on for its video platform. It connects "approved AI tools like Claude and ChatGPT to your video library," and Panopto names Microsoft Copilot too. Anthropic's Claude for Education update (9 July 2025) first announced Panopto and Wiley as pre-built MCP servers.

Best for: Universities and companies with large recorded-lecture or training libraries.

Pricing: Sold as a platform add-on. Panopto lists no price; contact its team.

Pros:

• Panopto says the server "is read-only by design," so the AI cannot change, delete or create content.
• Existing Panopto permissions decide what each user can see.
• Lets students search lectures from inside the AI app they already use.

Cons:

• An add-on with no public price, so budget is unclear until you ask.
• Only as useful as your transcripts and captions.

Checked against: Panopto MCP Server page on 2 November 2026.

3. Wiley Scholar Gateway: peer-reviewed sources for AI answers

What it is. Wiley's Scholar Gateway connector "enables Claude to generate responses grounded in peer-reviewed sources." It searches Wiley journals and returns source metadata through two tools, search_wiley_fulltext and getUsageLimit. The listing names John Wiley & Sons Inc. as publisher.

Best for: Students, faculty and librarians who want citations from Wiley content.

Pricing: Sign-in is required. Wiley publishes no price; the Scholar Gateway docs list scholargateway@wiley.com for product support.

Pros:

• Answers are grounded in peer-reviewed journal articles.
• Source metadata makes citations easy to check.
• A usage-limit tool helps you see remaining allowance.

Cons:

• Wiley content only, so it is one publisher's slice of the literature.
• Access terms and pricing are not published.

Checked against: Claude connector listing on 2 November 2026.

4. Learning Commons Knowledge Graph: standards and learning progressions

What it is. The Knowledge Graph MCP server from Learning Commons runs at https://kg.mcp.learningcommons.org/mcp. Tools such as find_standard_statement, find_learning_components_from_standard and find_standards_progression_from_standard look up US academic standards, math learning components and prerequisite standards. Some tools, such as misconceptions and lesson materials, need extra dataset access.

Best for: Curriculum teams and edtech companies aligning content to state standards.

Pricing: Not published. You need an API key from the Learning Commons Platform.

Pros:

• Standards lookups default to Common Core, with state-specific options.
• Math progressions show what comes before and after a standard.
• Your agent only sees the tools your data access allows.

Cons:

• The docs call it early access and say breaking changes may occur.
• Coverage is uneven: progressions are math only, and learning components cover math and K-2 ELA.

Checked against: Learning Commons MCP docs on 2 November 2026.

5. Open edX MCP: admin and authoring tools with guardrails

What it is. Open edX MCP, announced on the Open edX blog on 24 September 2026, lets AI assistants run a defined set of admin and course-authoring tasks. It was built by Blend-ed and ships as a Django app plus a Tutor plugin. Tools cover enrollments, course roles, Studio content, certificates, grade reports and analytics.

Best for: Universities and training providers that run their own Open edX instance.

Pricing: Free and open source under AGPL-3.0, per the GitHub repository. You pay for your own hosting.

Pros:

• High-impact actions return a dry-run preview and need a single-use confirmation token.
• Least-privilege scopes, per-tool rate limits and an append-only audit log of writes.
• Keys belong to an existing staff account and can be revoked.

Cons:

• A first release targeting the Ulmo version, with no tagged releases on GitHub yet.
• The README warns rate limits multiply per worker unless you use a shared cache such as Redis.

Checked against: Open edX blog post on 2 November 2026.

6. Teachfloor: a hosted LMS with its own MCP server

What it is. Teachfloor is a learning platform for cohort-based courses and training. Its MCP server page describes read and write access to courses, modules, learners, cohorts, enrollments, certificates, automations and analytics, mirroring the Teachfloor API. Sign-in uses OAuth 2.0 with your Teachfloor account.

Best for: Training companies and course creators running cohorts on Teachfloor.

Pricing: The Teachfloor pricing page starts at $89 per month (Startup), but lists the MCP Server under the custom-quote Full Features plan.

Pros:

• Actions run with the signed-in user's role, and every call goes to the audit log.
• Admins can disable specific tools per workspace.
• Teachfloor names Claude, ChatGPT, Gemini, Copilot, Cursor and Mistral as supported clients.

Cons:

• Not available on the entry plan.
• Only useful if Teachfloor is your LMS.

Checked against: Teachfloor MCP server page on 2 November 2026.

7. Canvas MCP (community-built): the most complete Canvas option

What it is. Canvas MCP is a community-built server by Vishal Sachdev, not by Instructure. It has read-only tools for students (assignments, grades, to-dos) and educator tools for assignments, grading, modules, pages and messaging. The README describes "up to 128 tools," while the repository summary says "up to 102," depending on which features are switched on.

Best for: Canvas instructors and course designers comfortable with a local setup.

Pricing: Free and open source under the MIT license.

Pros:

• Optional anonymization replaces student identity fields with consistent anonymous IDs.
• Log redaction and email masking are built in, and the project says it adds no telemetry.
• Student tools are read-only, and inbox messaging is off by default.

Cons:

• Not supported by Instructure; it uses your personal Canvas API token, which many schools issue only through IT.
• Educator tools include bulk grading, so a mistake can reach many students at once.

Checked against: Canvas MCP GitHub repository on 2 November 2026.

8. Microsoft Learn MCP Server: free docs for technical training

What it is. The Microsoft Learn MCP Server is a remote server at https://learn.microsoft.com/api/mcp. It lets agents search Microsoft's official documentation, fetch full articles and search code samples. Microsoft says it covers public documentation, "not training or user profile information."

Best for: IT training teams, bootcamps and learners studying Microsoft and Azure skills.

Pricing: Free. Microsoft says "There's no charge to use the MCP server" and "There's no authentication required."

Pros:

• No sign-in and no cost.
• Content refreshes daily from official docs.
• Works in VS Code, Visual Studio, Foundry and other MCP clients.

Cons:

• It covers Microsoft documentation only, not learner progress or course data.
• No access to Microsoft Learn training records or certifications.

Checked against: Microsoft Learn MCP overview on 2 November 2026. Related profile: Microsoft MCP.

Student privacy comes first

FERPA "is a federal law that affords parents the right to have access to their children's education records," according to the US Department of Education's student privacy office. Connecting an AI app to a gradebook moves those records into a new system, so treat it like any other vendor review.

1. Start with read-only tools. Panopto, Microsoft Learn and Canvas MCP's student tools only read.
2. Anonymize where you can. Canvas MCP's anonymization mode is a good model.
3. Keep writes behind a person. Grades, announcements and enrollments should need human approval. Open edX's dry-run and confirmation tokens show how.
4. Check the scopes. Use our MCP tool permission planner to map which tools each role should get.

Our guide to excessive permissions in MCP explains why broad tokens are the biggest risk.

What we left out

• Instructure Canvas, Blackboard, D2L Brightspace and Moodle. We found community servers and third-party mentions, but no vendor-published MCP documentation we could open. Anthropic's education post describes a Canvas LTI integration, which is not an MCP server.
• Khan Academy, Coursera and Quizlet. We found only unofficial projects or nothing at all.

Next step

If your LMS or student information system has no MCP server, or you need one with stricter scopes and audit logs than the community options, our MCP development team builds custom servers for education platforms.

Frequently asked questions