How karakeep-app/karakeep works
Karakeep is a self-hostable bookmark manager: users save links, notes, images, and PDFs, then organize, search, read, and revisit them across web, mobile, browser, and agent-oriented clients. The documented product also includes crawling and archival, AI tagging and summaries, semantic search, RSS, rules, highlights, imports, collaboration, and offline mobile reading. The graph shows these major capabilities while keeping asynchronous processing distinct. Some backend modules and storage implementations are not present in the supplied tree or excerpts, so uncertain internal connections are omitted.
Main entry points: route.ts API and tRPC · app Mobile app · assetStorage.ts Crawling and archival · inferenceWorker.ts AI tagging and summaries
Components 25
User clients
- Web app
apps/web/app - Mobile app
apps/mobile/app - Browser extension
apps/browser-extension/src - CLI
apps/cli/src - MCP tools
apps/mcp/src
Bookmark experience
- Bookmark views
apps/web/app/dashboard/bookmarks/page.tsx - Reader and highlights
apps/web/app/reader/[bookmarkId]/page.tsx - Full-text and semantic search
apps/web/app/dashboard/search/page.tsx
Background processing
- Crawling and archival
apps/workers/workers/crawler/assetStorage.ts - AI tagging and summaries
apps/workers/workers/inference/inferenceWorker.ts - Embedding indexing
apps/workers/workers/embeddingsWorker.ts - Rule execution
apps/workers/workers/ruleEngineWorker.ts
Organization and access
- Lists and sharing
apps/web/app/dashboard/lists/page.tsx - Tags and favorites
apps/web/app/dashboard/tags/page.tsx - RSS feeds
apps/web/app/settings/feeds/page.tsx - Rule configuration
apps/web/components/dashboard/rules/RuleEngineRuleEditor.tsx - Import sessions
packages/trpc/models/importSessions.service.ts - Rule engine
packages/trpc/lib/ruleEngine.ts
Data and integrations
- API and tRPC
apps/web/app/api/[[...route]]/route.ts - Application database
- Job queues
- Vector search store
- Inference provider
- Meilisearch
Other components
- User
Connections 35
- User → Web app: uses
- User → Mobile app: uses
- User → Browser extension: saves with
- User → CLI: uses
- User → MCP tools: asks agents
- Web app → API and tRPC: requests
- Mobile app → API and tRPC: requests
- Browser extension → API and tRPC: saves bookmarks
- CLI → API and tRPC: commands
- MCP tools → API and tRPC: invokes tools
- API and tRPC → Application database: reads and writes
- API and tRPC → Job queues: enqueues work
- Bookmark views → API and tRPC: loads and updates
- Reader and highlights → API and tRPC: loads content
- Full-text and semantic search → API and tRPC: queries
- Lists and sharing → API and tRPC: manages lists
- Tags and favorites → API and tRPC: manages tags
- RSS feeds → API and tRPC: configures feeds
- Rule configuration → API and tRPC: saves rules
- Import sessions → Application database: stages imports
- Job queues → Crawling and archival: dispatches jobs
- Job queues → AI tagging and summaries: dispatches jobs
- Job queues → Embedding indexing: dispatches jobs
- Job queues → Rule execution: dispatches jobs
- Crawling and archival → Application database: stores crawl data
- Crawling and archival → Job queues: enqueues follow-up
- AI tagging and summaries → Application database: reads and updates
- AI tagging and summaries → Inference provider: requests inference
- Embedding indexing → Application database: loads bookmark data
- Embedding indexing → Vector search store: writes vectors
- Rule execution → Rule engine: runs rules
- Rule engine → Application database: reads and updates
- Rule engine → Job queues: queues archival
- Full-text and semantic search → Meilisearch: searches content
- Mobile app → Reader and highlights: reads offline