Karakeep

Karakeep is an open-source, self-hosted bookmark manager and read-it-later hub that allows users to save links, notes, images, and PDFs in a single location. The software is deployed as a server-side application and is accessible via a web interface, mobile apps for iOS and Android, and browser extensions for Chrome, Firefox, and Safari.

The system focuses on data collection and organization, utilizing automation to reduce manual effort. It includes tools for archiving content to prevent link rot and a variety of importers to migrate data from other bookmarking services such as Linkwarden. Users can manage their collections through lists and collaborate with others on shared content, making it a viable tool for both individual research and team knowledge bases.

Key features

• LLM-based automatic tagging and summarization with support for local models via Ollama
• Full page archival using monolith and video archiving via yt-dlp
• Full text search across all stored content using Meilisearch
• OCR for extracting text from saved images
• Auto hoarding of content from RSS feeds
• Rule-based engine for customized content management
• Support for highlights and marks within hoarded content
• REST API for integration with external clients and LLM agents
• Importers for Chrome, Linkwarden, and Tab Session Manager

Built with NextJS, Drizzle, and tRPC, Karakeep is designed for users who prefer to own their data and avoid proprietary cloud lock-in. It integrates with Floccus for automatic synchronization with browser bookmarks and supports SSO for authentication. The architecture employs Puppeteer for crawling bookmarks and NextAuth for user management. This technical stack allows the application to handle a wide variety of media types, making it suitable for researchers, developers, and digital hoarders who require a centralized repository for miscellaneous web fragments.

Karakeep positions itself as a self-hosting first tool for comprehensive digital archiving and knowledge management.