AoA Course Connector
aoa-course-connector
An MCP server that reads courses from GetCourse, Skillspace and Stepik and turns them into a local, searchable knowledge base with citations
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Uses the user's saved browser session to access the course
- Stores course content locally outside the platform's control
Install
Manual install
pip install -e ".[mcp]"Install from a cloned repository; there is no PyPI release.
This is third-party code. Review the repository files before installing.
What it does
The tool pulls course content the user already has access to via a browser session (GetCourse, Skillspace) or an open API (Stepik) and normalizes it into a common structure: modules, lessons, steps, transcripts, assignments. It builds a local index for keyword and semantic search plus a graph of links between materials. The MCP server exposes read, search and answer tools that carry the source, capture time and freshness of every fact. A separate aoa-course CLI covers the same operations without an agent, along with local integrity checks for the index.
Who it is for. For people taking or running courses on GetCourse, Skillspace or Stepik who want an agent to search their own course materials.
Good fit when
- You need to quickly find a lesson or transcript fragment in a large course
- You want local search over your own courses without sending course content to an external AI service
- You need to verify the course index has not drifted from the source
Not a fit when
- You want a ready product with no setup: you need your own browser session and platform login
- Your course is not on GetCourse, Skillspace or Stepik: other platforms are only planned
- You want to pull someone else's paid course without access: the tool only works with courses you already have legitimate access to
Example request
Find the lesson in my GetCourse course that explains the sales funnel and quote the relevant fragmentLimitations
The project has a single author, there are no PyPI releases, and it installs only from source. GetCourse and Skillspace need a saved, authorized browser session, and the README does not spell out the steps to obtain one. Semantic search is local and simple by default; better quality needs your own HTTP embedding service. Moodle, Canvas, Coursera and other platforms are listed as planned, not working adapters.
How to disable. Remove the aoa-course-connector-mcp server from your MCP client config and uninstall the Python package from your environment if needed.
MCP
- Transport
- stdio
- Authentication
- not required
| Environment variables | |
|---|---|
| AOA_COURSE_DATA_ROOT | Directory for local course data outside Git, referenced in the README as AOA_COURSE_* variables. |
Security check
- Uses the user's saved browser session to access the course
- Stores course content locally outside the platform's control
README in short
The README lays out the architecture in detail: source adapters, normalization into shared course objects, local indexes and a graph, citation-backed search, and an MCP server with a CLI twin. It separately notes fixture sources used only to demonstrate the method, and draws a line to runtime deployment and quality evaluation, which live in the same author's other projects. Apache-2.0 license.
FAQ
Can it download someone else's paid course without access?
No. The tool only works with courses the user already has legitimate access to and describes itself as permission-aware.
Does search need an external AI service?
No, it uses a local deterministic search by default. An external HTTP embedding service can be added optionally for better semantic search.
Related
A self-hosted knowledge base with block-level references and a built-in MCP server for connecting AI agents to your notes
A CLI for every Google Workspace API with JSON output and agent skills: Drive, Gmail, Calendar, Sheets and more
Local search over Markdown notes, docs and meeting transcripts: keywords, semantic search and reranking, with an MCP server
A task manager for AI-driven development: breaks a PRD into dependent tasks and guides the agent through them via MCP or CLI