claudedwithlove
explore/howcani

howcani

Cherished

HowCanI is a self-hosted FAQ and knowledge base application that lets you store, search, and organize personal documentation locally. Built with Bun and Elysia on the backend and Svelte 5 on the frontend, it combines full-text search with optional AI-powered semantic search via vector embeddings, accessible securely over WireGuard VPN.

·0··submitted April 16, 2026
View on GitHub
Clauded With Love Rating
7.6 / 10

HowCanI is a self-hosted personal knowledge base application combining traditional full-text search with AI-powered semantic search via vector embeddings. It features a modern tech stack (Bun/Elysia backend, Svelte 5 frontend) with comprehensive deployment options including Docker, VPN access, and MCP integration for AI clients.

Code Quality7.5
Usefulness8.2
Claude Usage7.0
Documentation8.5
Originality6.8
Highlights
  • Hybrid search implementation using Reciprocal Rank Fusion to combine FTS5 and vector embeddings provides sophisticated search capabilities
  • Comprehensive deployment documentation with Docker scripts, backup procedures, and VPN configuration shows production-ready thinking
  • MCP server integration enables AI clients like Claude to interact with the knowledge base, demonstrating forward-thinking AI tooling integration
To Improve
  • Add test coverage for the hybrid search implementation and vector embedding pipeline to ensure reliability of the core AI features
  • Implement rate limiting and input validation for the MCP endpoints to prevent abuse and ensure security when exposed to AI clients
Topic
Stack