Stars
Forks
Watchers
Developer links
Blinko
With over 10,800 GitHub stars earned in under two years, Blinko has emerged as the privacy-first answer to cloud-dependent note-taking tools by combining instant thought capture with AI-powered retrieval that actually understands what you wrote rather than just matching keywords. The core engine uses Retrieval-Augmented Generation to build vector embeddings of every note in your PostgreSQL database, enabling natural language queries like "what were my thoughts on the database migration last Tuesday" to surface relevant content through semantic understanding rather than exact string matching. AI integration supports both cloud providers — OpenAI, MiniMax, and compatible endpoints — and fully local inference through Ollama running models like Llama 3.2 on your own hardware, ensuring your notes never leave your network when privacy demands it. Built on Next.js with a React frontend, the web interface presents notes as cards with full Markdown support including code blocks, LaTeX, and rich formatting, with a clean input bar for capturing fleeting thoughts in seconds. The Tauri-based desktop and mobile clients extend access to macOS, Windows, Linux, and Android with native performance characteristics. Notes are stored as plain text in PostgreSQL with vector indexes that update incrementally as new content arrives, and the embedding index can be rebuilt on demand when switching between AI providers. Multiple users can share a single instance with individual accounts, and data exports to standard formats. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. GPL-3.0 licensed.
Benefits
- AI That Understands Context
- RAG-powered semantic search using vector embeddings finds notes by meaning rather than keywords, supporting natural language queries across your entire note library instantly.
- Complete Privacy Control
- Run AI inference locally via Ollama with models like Llama 3.2 so notes never leave your server, or use cloud providers when convenience outweighs privacy concerns.
- Instant Thought Capture
- Card-based interface with a quick-input bar captures fleeting ideas in seconds with full Markdown support including code blocks, LaTeX, and rich text formatting.
- Cross-Platform Native Apps
- Tauri-based desktop and mobile clients for macOS, Windows, Linux, and Android deliver native performance while syncing with your self-hosted PostgreSQL instance.
Features
- RAG Note Retrieval
- Vector embedding search powered by Retrieval-Augmented Generation enables semantic queries using OpenAI, Ollama, MiniMax, or compatible AI provider endpoints.
- Markdown Card Notes
- Capture thoughts as plain-text cards with full Markdown support including code syntax highlighting, LaTeX equations, and structured formatting.
- Local AI Inference
- Connect Ollama for fully on-premise AI processing where notes never leave your hardware, supporting models like Llama 3.2 without external API dependencies.
- Multi-Platform Tauri Clients
- Native desktop and mobile applications built with Tauri for macOS, Windows, Linux, and Android with lightweight resource usage and responsive performance.
- Incremental Vector Indexing
- PostgreSQL-stored vector embeddings update automatically as new notes arrive, with on-demand full index rebuilds when switching between AI providers.