14 core query tools
Course catalogs, completion reports, enrollments, assignments, activity feeds, category navigation, user lookups, and more.
moodle-mcp-serverlets any AI agent query a Moodle instance in plain English and get structured data back -- no report builders, no SQL, no CSV exports. It's free, open source, and read-only, built to drop into a real Moodle environment without modifying it.
moodle-mcp-serveris a Model Context Protocol server that connects AI agents directly to Moodle's Web Services API. Instead of learning Moodle's reporting tools, writing queries, or exporting spreadsheets, you ask questions in plain language and the agent returns the answer as structured data.
It runs as a local process, talks to the model of your choice, and never writes to your Moodle.
Course catalogs, completion reports, enrollments, assignments, activity feeds, category navigation, user lookups, and more.
Use Claude, GPT, Gemini, or a local model through Ollama or llama.cpp. Nothing is locked to a single provider.
It queries Moodle through the Web Services API and never modifies LMS data.
Enable Web Services, create an API token, and connect the server. There is no Moodle-side plugin install.
Pair it with a local model and keep the whole workflow on hardware you control.
Add new tool modules through the documented runtime plugin contract as the reporting surface grows.
It's a clean fit for schools, colleges, training organizations, and LMS operators that need answers out of Moodle but care about where the data goes.
The core is free and open source. Advanced reporting, charting, analytics, and multi-instance management are being built as commercial extensions on top of it for teams that want more than raw queries. The foundation is open; the deeper capabilities are optional.
The repository includes the server, reference client, setup instructions, and plugin docs. Start with the Quick Start if you want the shortest path into a working local install.
These answers cover the practical setup and data-handling questions teams usually ask first.
No. moodle-mcp-server is read-only. It queries through Moodle's Web Services API and never writes anything back, so it's safe to run against a live production instance.
No. It works with Moodle's existing Web Services API. You enable Web Services and create an API token. There is no plugin to install on the LMS itself.
Any MCP-compatible client or model path. That includes cloud providers like Claude, GPT, and Gemini, as well as local models run through Ollama or llama.cpp.
Yes. Paired with a local model, the stack can run on your own hardware so Moodle data stays inside your network, which matters for organizations with privacy or compliance requirements.
The core is self-serve and open source. If you'd rather have help setting it up -- or you're an institution thinking through privacy, hosting, and integration -- tell us what you're working with.