A fully local RAG pipeline using LlamaIndex + Ollama + Chroma to query your Logseq notes.

• Python 3.13+

• Ollama running (https://ollama.com)

• Pull a chat and embedding model:

  ollama pull llama3.1
  ollama pull nomic-embed-text

  or (lighter weight):

  ollama pull llama3.1
  ollama pull all-minilm

cd logseq-chat
make install

Edit config.yaml and at a minimum set logseq_root to your Logseq graph directory.

make ingest

make chat

make test

• Summarize tasks tagged #home in October 2025.
• Find notes referencing [[Team Topologies]] and list my pros/cons.

• Skips assets/ by default. Enable OCR later if needed.
• Uses Markdown-aware chunking; tags from #tag and tags:: stored in metadata.
• For faster machines, try bigger models; for CPU-only, consider llama3.2 or qwen2.5:7b and smaller chunks.