Documentation

Docker Setup

For Personal, install the app with the download zip (a1kh, setup.sh, run.sh). Use Docker when you want Ollama — and optionally Docling-serve — in containers instead of installing them on the host.

Requires Docker Engine 20.10+ and Docker Compose 2.0+.

Personal uses sqlite-vec inside the app process. There is no Qdrant (or other vector DB) container.

Recommended layout

  1. Start Ollama (and Docling if needed) with Compose
  2. Unzip the Personal kit on the host
  3. Point config.yaml at http://localhost:11434 (Ollama) and, if used, Docling on the published port
  4. Run ./setup.sh then ./run.sh

Docker Compose (dependencies only)

services:
  ollama:
    image: ollama/ollama:latest
    container_name: ollama
    ports:
      - "11434:11434"
    volumes:
      - ./ollama_data:/root/.ollama
    restart: unless-stopped

  # Optional — only if importer.provider is docling
  docling:
    image: quay.io/docling-project/docling-serve:latest
    container_name: docling
    environment:
      DOCLING_SERVE_ENABLE_UI: "1"
    ports:
      - "5001:5001"
    restart: unless-stopped

Save as docker-compose.yaml and start:

docker compose up -d

Pull models into the Ollama container

docker compose exec ollama ollama pull lfm2.5:latest
docker compose exec ollama ollama pull embeddinggemma-2:270m
docker compose exec ollama ollama list

Host config.yaml URLs

With ports published to the host (as above):

llm:
  provider: ollama
  url: "http://localhost:11434"

# Only when using Docling:
importer:
  provider: docling
  url: "http://localhost:5001"

Without Docling, keep importer.provider: none (shipped default).

Day-to-day commands

docker compose up -d
docker compose logs -f ollama
docker compose stop
docker compose down          # keep volumes
docker compose down -v       # deletes ollama_data — models gone

Port conflicts

If 11434 or 5001 is taken, change the host side of the mapping, for example "11435:11434", and update llm.url / importer.url to match.

Troubleshooting

  • App cannot reach Ollama: confirm docker compose ps, curl http://localhost:11434/api/tags, and llm.url in config.yaml
  • Docling configured but serve fails: either start Docling or set importer.provider: none
  • Volume permissions (Linux): sudo chown -R $USER:$USER ./ollama_data

Next steps