# MonMaster AI

> Serveur MCP connectant les assistants IA à la base officielle des masters français (7 293 formations)
> MCP server connecting AI assistants to the official French master's degree database (7,293 programmes)

MonMaster AI is an open-source Model Context Protocol (MCP) server that gives AI assistants
real-time access to the official French master's degree database from monmaster.gouv.fr.

## What it does

- Semantic search across 7,293 French master's degree programmes using vector embeddings (BAAI/bge-m3, 1024 dimensions)
- Personalised recommendations based on field of study, location, and employment rates
- Detailed programme profiles: contacts, required documents, tuition fees, language of instruction
- Acceptance statistics: selection rates, number of applications, rankings
- Aggregate statistics by discipline (mention)

## MCP Server

- SSE Endpoint: https://monmaster-ai.com/sse
- Protocol: Model Context Protocol (MCP) by Anthropic
- Transport: SSE (Server-Sent Events)
- Authentication: None required (public server)

## REST API

- Base URL: https://monmaster-ai.com/api
- OpenAPI schema: https://monmaster-ai.com/api/openapi.json
- Interactive docs: https://monmaster-ai.com/api/docs

## Available MCP Tools

### list_formations(mention?: string) -> Formation[]
List all master's programmes, optionally filtered by discipline (mention).
Example: list_formations(mention="Informatique")

### get_formation(id_or_url: string) -> FormationDetail
Get full details of a programme by its numeric ID or URL slug.
Example: get_formation(id_or_url="42")

### search_formations(query: string, mention?: string) -> Formation[]
Keyword search across programme names, descriptions, and disciplines.
Returns: access_rate, language, fees, establishment, mention.
Example: search_formations(query="intelligence artificielle Paris")

### get_stats(mention?: string) -> Stats
Aggregate statistics: total count, access rate distribution, top establishments.
Optionally scoped to a discipline.
Example: get_stats(mention="Droit")

### recommend_formations(query: string, limit?: int) -> Formation[]
Semantic recommendations using vector similarity + employment rate scoring.
Example: recommend_formations(query="data science machine learning Lyon")

## Data Model

Each formation contains:
- id, formation_url (unique identifier)
- name, establishment, mention, parcours
- capacity (integer), access_rate (text, e.g. "15 %")
- fees, language, last_rank, total_applications
- contacts (JSON), required_documents (JSON)
- employment_stats (JSON), m2_pathways (JSON)
- embedding (vector, 1024 dimensions)

## Data Source

Official data scraped from monmaster.gouv.fr
(French Ministry of Higher Education and Research)
Last updated: 2026

## Technology Stack

- MCP Server: Python, FastMCP 2.x
- REST API: Python, FastAPI
- Database: PostgreSQL 16 + pgvector 0.8
- Embeddings: BAAI/bge-m3 via OpenRouter API (1024 dimensions)
- Infrastructure: Hetzner CX23, Debian 13

## Source Code

https://github.com/okyalos76/monMaster

## Website

- French: https://monmaster-ai.com
- English: https://monmaster-ai.com/en

## Contact

Project maintained by okyalos76. Issues and contributions welcome on GitHub.