Taming Your Bookmarks: Why I Chose Self-Hosted KaraKeep

We all know the struggle: hundreds of bookmarks piled up in Chrome or Firefox, tucked away in folders we never open. Let’s be honest, default browser management is insufficient. It lacks efficient full-text search, smart categorization, and most importantly, it offers zero visibility into the actual content of the link.

This is where KaraKeep comes in. I decided to self-host it to take back control of my bookmarks while adding a layer of artificial intelligence.

Why KaraKeep?

Unlike traditional solutions, KaraKeep does more than just store a URL. It captures the content, enables ultra-fast searching thanks to Meilisearch, and, best of all, integrates with OpenAI to automatically tag and summarize your readings.

Architecture and Docker Compose Installation

The setup relies on three pillars:

1. KaraKeep (Web): The user interface.
2. Meilisearch: The search engine.
3. Alpine Chrome: A headless browser that allows KaraKeep to “read” pages for indexing.

Here is the docker-compose.yml file I use for my deployment (configured to run behind a Traefik reverse proxy):

services:
  web:
    image: ghcr.io/karakeep-app/karakeep:${KARAKEEP_VERSION:-release}
    restart: unless-stopped
    volumes:
      - data:/data
    env_file:
      - .env
    networks:
      - web
      - back
    labels:
      - traefik.enable=true
      - traefik.docker.network=web
      - traefik.http.routers.books.entrypoints=websecure
      - traefik.http.routers.books.tls.certresolver=myresolver
      - traefik.http.routers.books.rule=Host(`bookmarks.bracloud.fr`)
      - traefik.http.services.books.loadbalancer.server.port=3000

  chrome:
    image: gcr.io/zenika-hub/alpine-chrome:124
    restart: unless-stopped
    command:
      - --no-sandbox
      - --disable-gpu
      - --disable-dev-shm-usage
      - --remote-debugging-address=0.0.0.0
      - --remote-debugging-port=9222
      - --hide-scrollbars
    networks:
      - back

  meilisearch:
    image: getmeili/meilisearch:v1.13.3
    restart: unless-stopped
    env_file:
      - .env
    environment:
      MEILI_NO_ANALYTICS: "true"
    volumes:
      - meilisearch:/meili_data
    networks:
      - back

networks:
  web:
    external: true
  back:
    external: true

volumes:
  meilisearch:
  data:

.env File Configuration

The .env file is the heart of the configuration. This is where you link the services and activate the AI features.

🔑 Core Variables

These variables allow the web container to communicate with the search engine and the browser.

• KARAKEEP_VERSION: Defines the image version (e.g., release).
• MEILI_MASTER_KEY: A complex key to secure your search engine.
• SECRET_KEY: Used to secure sessions (generate one via openssl rand -hex 32).

🤖 OpenAI: Turbocharging Your Bookmarks

This is where the AI magic happens. Once enabled, OpenAI will analyze every link to create relevant summaries and tags.

• OPENAI_API_KEY: Your secret key retrieved from platform.openai.com.
• OPENAI_MODEL: Use gpt-4o-mini for the best price/performance ratio (OpenAI pricing).
• AUTO_AI_SUMMARY: Set to true for automation.
• AUTO_AI_TAGS: Set to true for automatic keyword extraction.

Full .env Example:

# --- GENERAL ---
KARAKEEP_VERSION=release
SECRET_KEY=your_random_key
DATA_DIR=/data

# --- MEILISEARCH ---
MEILI_ADDR=http://meilisearch:7700
MEILI_MASTER_KEY=SuperSecureKey123!

# --- BROWSER (SCRAPING) ---
BROWSER_WEB_URL=http://chrome:9222

# --- ARTIFICIAL INTELLIGENCE ---
OPENAI_API_KEY=sk-proj-XXXX...
OPENAI_MODEL=gpt-4o-mini
AUTO_AI_SUMMARY=true
AUTO_AI_TAGS=true

Conclusion

Moving from “browser-based” management to a self-hosted KaraKeep instance radically changes how you consume information. Thanks to Meilisearch’s full-text indexing and OpenAI’s intelligence, your bookmarks are no longer a dead list, but a true knowledge base.