italoo97/study-mcp

MCP server to ingest and search study materials using semantic search

About study-mcp

A Model Context Protocol (MCP) server that turns video transcripts and study materials into a searchable, AI-powered knowledge base — directly inside Claude Desktop.

Paste a video transcript or point to a PDF, slide deck, DOCX, or image and instantly:

- 🎬 Ingest video transcripts — paste raw text or load .srt/.vtt caption files; timestamps are preserved so you can jump back to the exact moment in the video - 🔍 Search semantically — ask questions in natural language across all your materials, powered by local embeddings (no API cost) - 🎯 Generate quizzes — sample representative context from any material so Claude can quiz you on it - 🧠 Summarize & create flashcards — saved directly to Notion - 📊 Track your library — list materials, inspect overviews, view stats, delete what you no longer need

Everything runs locally by default (ChromaDB + HuggingFace embeddings). Set a single environment variable to switch to pgvector on Supabase/Postgres.

Layer Technology ------ Language Python 3.13 MCP server MCP Python SDK (FastMCP) Embeddings sentence-transformers + HuggingFace models (local inference, PyTorch) Vector stores ChromaDB (local) · pgvector on Supabase/Postgres (cloud) Document parsing Docling (PDF, DOCX, PPTX, HTML, images) + native SRT/VTT parser Integrations Notion API (notion-client) Configuration pydantic-settings (typed, env-based) Tooling Poetry · pytest (+coverage) · ruff · mypy --strict · pre-commit · commitizen CI GitHub Actions (lint, type check, tests on every push/PR)

From the project README.

Maintaining this server?

Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:

[![On MCP Radar](https://mcp.liqiwa.com/badge.svg)](https://mcp.liqiwa.com/s/italoo97--study-mcp.html)

Related MCP servers