Cache: open-source bookmark manager unifies saves across platforms

Cache is an open-source bookmark manager designed to consolidate saves from multiple platforms—including Browser bookmarks, Instagram Saved, TikTok Favorites, YouTube Watch Later, X/Twitter bookmarks, GitHub Stars, Pinterest, and Google Photos—into a single unified library. The core insight driving the project is that bookmarking is fundamentally broken: when users deliberately save content, that intent immediately disappears into platform-specific lists they never revisit, scattered across services with no connection to their actual workflow or goals.

The tool operates around several core features. Smart collections automatically organize entries with AI-assisted relevance ranking that learns user preferences over time. Overviews provide a one-line summary above each collection that updates instantly as new entries arrive. An AI-assisted search feature lets users ask a Cache agent to search across all saved content. The platform also supports first-party note-taking alongside bookmarks, live collaboration (shareable collection views that work even for non-Cache users), custom automations for generating daily digests and reminders, and a browser extension for capturing content from anywhere on the web. Importantly, Cache imposes no limits on the number of saves, unlike competing tools.

Cache can be used as a cloud-hosted service at www.cachd.app, or self-hosted for complete data control. Self-hosting requires Bun v1.3.14+, Node.js 24.x, PostgreSQL 12+, and a Google Gemini API key for AI features. The project is built on Next.js with React and Tailwind CSS for the frontend, PostgreSQL and Redis for the database, Better Auth for authentication, AI SDK and Gemini for AI/LLM capabilities, and Stripe for subscriptions. The cloud version includes Arcjet for security (WAF, rate limiting, PII redaction).

A notable feature is the MCP (Model Context Protocol) server exposed at https://www.cachd.app/mcp, which allows AI agents like Claude and Cursor to integrate directly with a user's Cache library, enabling them to search bookmarks, save new items, list collections, and perform other actions without leaving the agent interface. The upcoming roadmap includes reminders, threaded comments on entries, an inbox triage view for reviewing saves, richer note-taking with advanced formatting, Raycast integration for capture and search, and Substack integration for importing and saving posts and newsletters. The project is licensed under Apache License 2.0 and welcomes community contributions.

The core problem Cache addresses is that bookmarks—an intentional signal that content deserves remembering—typically become lost in platform-specific lists that users never revisit. The fragmentation across browsers, social platforms, and cloud services disconnects saving from workflow and discovery. Existing tools treat bookmarking as a low-priority feature rather than the starting point for content management.

Cache's design philosophy centers on treating the save action as a first-class event. By unifying bookmarks across multiple platforms and pairing them with AI-assisted search, automatic organization, and live summaries, the tool aims to transform a deliberate but wasted signal into actionable knowledge. The inclusion of note-taking, collaboration, and custom automations positions it as a workflow hub rather than a passive archive. The availability of self-hosting with zero telemetry by default appeals to users who prioritize data control and privacy, while the cloud option provides simplicity for those who prefer not to manage infrastructure.

No comments yet. Be the first to share your thoughts!