Karakeep Self-Hosted Bookmark Manager with AI Tagging
Karakeep (formerly Hoarder) is a self-hosted bookmark-everything app with 24K+ GitHub stars that provides AI-powered automatic tagging, full-text search, link archiving, and a REST API. It supports bookmarking links, notes, images, and PDFs with OCR extraction and RSS feed auto-hoarding.
npx skills add agentskillexchange/skills --skill karakeep-bookmark-manager-ai-tagging
Karakeep, previously known as Hoarder, is a self-hosted bookmark management application designed for data hoarders who want full control over their saved content. With over 24,000 GitHub stars and active development, it provides AI-powered automatic tagging, full-text search, and comprehensive content archiving through a modern web interface, mobile apps, and browser extensions.
What this skill actually does
Core Features
The application supports bookmarking links, taking notes, and storing images and PDFs. It automatically fetches link titles, descriptions, and preview images. Content can be organized into collaborative lists, and a full-text search engine powered by Meilisearch enables finding any stored content. OCR extraction pulls text from images, and RSS feed integration enables automatic content hoarding from specified sources.
AI-Powered Organization
Karakeep integrates with OpenAI or local Ollama models to provide automatic tagging and summarization of bookmarked content. A rule-based engine allows creating custom management rules for automated organization. The AI system analyzes content and assigns relevant tags without manual intervention, making it possible to organize large collections of saved content efficiently.
REST API and Integration
The application exposes a REST API that agents can use to programmatically create bookmarks, manage lists, search content, and retrieve tagged items. Browser extensions are available for Chrome and Firefox for quick bookmarking, and native iOS and Android apps provide mobile access. The platform supports SSO authentication, bookmark importing from Chrome, Pocket, Linkwarden, and Omnivore, and automatic sync with browser bookmarks via Floccus.
Architecture and Deployment
Built with Next.js, Drizzle ORM, tRPC, and NextAuth, Karakeep is designed as a self-hosting-first application. It uses Puppeteer for crawling bookmarked pages and Meilisearch for full-content search. Full page archival using Monolith protects against link rot, and automatic video archiving via yt-dlp captures video content. Deployment is straightforward via Docker Compose.
Agent Use Cases
Agents can use Karakeep to build and maintain knowledge bases from web research, automatically categorize and tag discovered resources, search across previously saved content for relevant information, archive important pages before they disappear, and manage curated collections of links and notes for research workflows.