Education MCP servers

MCP server for ICTExam, an AI exam authoring, delivery and auto-grading platform. Reads exam papers, classes, subjects, the teacher gradebook, and per-question item analysis (difficulty, p-value, average mark, full and zero credit counts). Read-only by default: parsing a question paper with AI, publishing and unpublishing are never registered with the model unless `ICTEXAM_MCP_ALLOW_WRITE=true`. Zero-install via `npx -y ictexam-mcp`.

Explainable AI-writing detection and originality checking for teachers, students and writers, in English **and Spanish**. `analyze_ai_writing` returns a 0–100 score *and every tell behind it* — overused vocabulary, rhetorical crutches, syntactic patterns and sentence-rhythm burstiness — each with the matched text and a concrete fix, instead of a black-box percentage. `check_originality` returns the shared passages between documents as evidence; `check_citations` cross-checks a bibliography against the text offline; `compare_to_baseline` runs Burrows's Delta against the same writer's earlier work; `inspect_characters` finds hidden and lookalike characters; `write_report` exports the evidence as a document a person can take to a committee. Every tool but two runs entirely on the machine and the text never leaves it; the two server-backed ones (perplexity, cross-language paraphrase) disclose that they send text. Published false-positive rate at its 30/100 boundary: 2 of 296 texts written before 2022 (0.7%, 95% interval 0.2–2.4%). .NET 10, MIT. Install: `dnx SignsOfAI.Mcp --yes`. Browser demo: <https://peopleworks.github.io/SignsofAI/>

Sudokumax connects AI assistants to eight Sudoku tools: generate verified unique puzzles, check uniqueness, analyze singles-based difficulty, validate progress, get hints, publish playable collections, export PDFs and create worksheet packs. Generate up to 20 puzzles with A4 or US Letter worksheets and answer sheets. Hosted Streamable HTTP with a free Bearer API key. Published output requires Sudokumax attribution. Documentation: https://sudokumax.com/developers/mcp. Contact: hello@sudokumax.com

Community pronunciation dictionary for developer project / product / jargon names (kubectl, nginx, GIF, JSON, Pydantic, Knative, LaTeX — 1,848 entries). Each entry carries IPA, a General-American respelling, alternate readings where the pronunciation is contested, a source citation (creator interview / project FAQ / Wikipedia § Pronunciation), and a confidence level. Tools: `pronounce(word)`, `search_pronunciations(query)`, `list_pronunciations(category)`. Install: `uvx pronounce-mcp`. Live browse: <https://pronounce.renlab.ai>.

Agent-native teaching OS: gives your AI a classroom that remembers. 50 tools for courses, lessons, FSRS flashcards, graded exercises, real-time teaching sessions, and teaching-memory read-back (`get_context`) — the agent authors and grades, reflects with evidence, and revises its own lessons; lesson completion belongs to the learner. Self-hosted (Postgres + web app), MIT.

Package HTML, Claude Design bundles or mobile-learning exports into LMS-ready SCORM 2004/1.2 courses — and validate any SCORM zip. Offline assets, milestone and question-level tracking, bundled ADL schemas.

Sudokumax connects AI assistants to eight Sudoku tools: generate verified unique puzzles, check uniqueness, analyze singles-based difficulty, validate progress, get hints, publish playable collections, export PDFs and create worksheet packs. Generate up to 20 puzzles with A4 or US Letter worksheets and answer sheets. Hosted Streamable HTTP with a free Bearer API key. Published output requires Sudokumax attribution. Documentation: https://sudokumax.com/developers/mcp. Contact: hello@sudokumax.com

MCP adapter for the Paxaver school community platform: student lunch ordering with allergy/dietary checks, wallets, school calendars, volunteering, memberships, and payments — 25+ tools over Streamable HTTP with OAuth 2.1 + PKCE and capability-first authorization. Remote `https://mcp.paxaver.com/mcp`, listed as a [hosted connector on Glama](https://glama.ai/mcp/connectors/com.paxaver.mcp/paxaver).

D2L Brightspace for students. Due dates with real submission status (never suggests work you already handed in), grades, assignment files and rubrics, content, announcements and discussions. Sign in through a normal browser, so it works at any school with SSO, Duo or MFA and needs no API key. `npx brightspace-d2l-mcp setup` connects 13 AI apps automatically.

MCP server for ICTExam, an AI exam authoring, delivery and auto-grading platform. Reads exam papers, classes, subjects, the teacher gradebook, and per-question item analysis (difficulty, p-value, average mark, full and zero credit counts). Read-only by default: parsing a question paper with AI, publishing and unpublishing are never registered with the model unless `ICTEXAM_MCP_ALLOW_WRITE=true`. Zero-install via `npx -y ictexam-mcp`.

Explainable AI-writing detection and originality checking for teachers, students and writers, in English **and Spanish**. `analyze_ai_writing` returns a 0–100 score *and every tell behind it* — overused vocabulary, rhetorical crutches, syntactic patterns and sentence-rhythm burstiness — each with the matched text and a concrete fix, instead of a black-box percentage. `check_originality` returns the shared passages between documents as evidence; `check_citations` cross-checks a bibliography against the text offline; `compare_to_baseline` runs Burrows's Delta against the same writer's earlier work; `inspect_characters` finds hidden and lookalike characters; `write_report` exports the evidence as a document a person can take to a committee. Every tool but two runs entirely on the machine and the text never leaves it; the two server-backed ones (perplexity, cross-language paraphrase) disclose that they send text. Published false-positive rate at its 30/100 boundary: 2 of 296 texts written before 2022 (0.7%, 95% interval 0.2–2.4%). .NET 10, MIT. Install: `dnx SignsOfAI.Mcp --yes`. Browser demo: <https://peopleworks.github.io/SignsofAI/>

Package HTML, Claude Design bundles or mobile-learning exports into LMS-ready SCORM 2004/1.2 courses — and validate any SCORM zip. Offline assets, milestone and question-level tracking, bundled ADL schemas.