RepoLens — AI GitHub Explorer
Built RepoLens, an AI-powered GitHub explorer that combines multi-dimensional repository search with LLM-driven code insights. Features include AI-generated What/How/Why project summaries, SSE streaming dialogue panel with RAG-enhanced context injection, trending rankings across 3 time windows and 12 languages, and full dark mode support. Compatible with any OpenAI-format AI service (Qwen, GLM, DeepSeek, Ollama).
Challenges
Integrating AI streaming (SSE) with real-time GitHub data for contextual Q&A
Efficiently handling GitHub API rate limits across search, details, and trending endpoints
Building a responsive UI that works across desktop sidebar and mobile bottom-sheet layouts
Solutions
RAG architecture: auto-injects README context into AI prompts for accurate codebase understanding
Server-side Octokit client with in-memory cache for deduplication and rate-limit optimization
Element Plus adaptive layout with keyboard shortcuts (⌘K/J/B) and mobile hamburger menu
Key Learnings
SSE streaming and RAG architecture for AI-powered applications
GitHub REST API optimization and caching strategies
Multi-model AI provider abstraction with OpenAI-compatible interface
Project Metrics
AI Models
4+ supported
Languages
12 filters