git clone https://github.com/keenanpepper/glowfic-rag.git
cd glowfic-rag

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

# Download the pre-built vector database (~2GB compressed, ~5GB on disk)
./setup_data.sh

claude

{
  "mcpServers": {
    "glowfic-search": {
      "command": "/path/to/glowfic-rag/.venv/bin/python3",
      "args": ["-m", "rag.mcp_server"],
      "cwd": "/path/to/glowfic-rag"
    }
  }
}

python -m rag.search "Keltham explains expected utility" -k 5
python -m rag.search "vampire meets werewolf" --continuity Sandboxes
python -m rag.search "dath ilan" --author Iarwain

# Scrape (supports --resume for interrupted runs)
python -m rag.scrape https://glowfic.com/boards/277

# Index (adds to existing collection, doesn't rebuild)
python -m rag.index data/trainwreck.jsonl

# Or reindex everything from scratch
python -m rag.index --reset