Fit-Wiki Scraper & MCP Server

Python tool for extracting course materials, exam variants, and notes from Fit-Wiki. Includes an MCP server for integration with LLM clients (Claude, AGY, OpenCode).

Requirements

• Python 3.12+

• Liberation Sans and Liberation Mono fonts (for correct Czech diacritics in PDFs)

Related MCP server: Deepwiki MCP Server

Setup

python3 -m venv venv
./venv/bin/pip install -r requirements.txt

Configuration

Copy .env.example to .env and fill in your cookies:

cp .env.example .env

Alternatively, pass FITWIKI_COOKIES as an environment variable in the MCP client config.

CLI Usage

Scrape a course to Markdown

./venv/bin/python scraper.py [course_code]

Convert Markdown to PDF

./venv/bin/python convert_to_pdf.py

Compiles all files in markdown_output/ to pdfs/.

Pipeline (single page)

./venv/bin/python index_page.py bi-osy            # download index
./venv/bin/python scraper.py                      # scrape to Markdown
./venv/bin/python convert_to_pdf.py               # compile PDFs

MCP Server

The MCP server exposes Fit-Wiki tools to LLM clients. It runs via stdio or SSE.

Client Configuration

Add the following to your client's MCP config file. Cookies are loaded from the .env file (see .env.example).

{
  "mcpServers": {
    "fitwiki": {
      "command": "/path/to/fitwiki/venv/bin/python",
      "args": ["/path/to/fitwiki/mcp_server.py"]
    }
  }
}

Config File Locations

Restart the client after saving. For AGY, verify with /mcp in chat.

Transports

• stdio (default) – used by the config above. The client spawns the server as a subprocess.

• SSE – for Docker or remote setups. Set MCP_TRANSPORT=sse and point the client to http://host:8000/sse.

Docker

Build & Run

docker compose up -d

The container exposes the MCP server on port 8000 via SSE. Mounted volumes persist markdown_output/ and pdfs/ on the host.

GitHub Container Registry

The Docker image is built automatically on push (see .github/workflows/docker-build.yml) and published to ghcr.io/<your-username>/fitwiki-mcp. Tags: latest, master, short SHA.

Exposed MCP Tools

• list_courses – List all courses on the FIT platform.

• list_course_sections – List sections (e.g. zkouska, test1) for a course.

• list_section_pages – List pages within specific sections.

• download_page – Scrape a page to Markdown and compile to PDF in one step.

• compile_pdf – Compile a Markdown file to PDF.

• compile_category_pdfs – Compile all Markdown files in a category to PDFs.

• scrape_index – Parse a course index and list discovered links.

• scrape_page – Scrape a single page to Markdown.

• scrape_course – Download all pages in selected categories for a course.