📚 background
In my Cubox I have thousands of bookmarks stored. Some bookmarks Cubox can successfully parse, grab content and title. However, for those bookmarks with vague information, if I don't enter the information manually, they will always be hidden in the corner that the search function cannot reach.
Cubox's full-text search feature is not always enough when we want to find a specific bookmark when we need it. This leaves us desperate for a better solution. That feeling of thinking of something but not being able to find it makes people jump anxiously.
🚀 CuboxGPT was born!
CuboxGPT uses GPT technology to help users better search for desired bookmarks in Cubox. It understands the semantics and thus finds the appropriate bookmark.
✨ Feature
- command line tool
- Python one-click installation
- Support OpenAI API Key
- local vector database
- The output contains citations and links
💬 How to use
- Export Cubox bookmark content
- Set OpenAI API environment variables locally
- Use cuboxgpt import-data to import bookmarks and grab webpage content
- Use cuboxgpt init-database to build a local vector database
- Natural language search with cuboxgpt search
⌛ Future plans
- Improve the accuracy of search content and make it easier for users to find the bookmarks they need
- Expand the number of search results and improve the search hit rate
- Improve database import update experience, support incremental update to reduce update cost
Visit Official Website
