asu-discover

Search ASU GitHub repositories by asking a natural language question. The system applies hybrid semantic and keyword retrieval to surface relevant integrations, SDKs, and best practices from 760+ indexed repos.

Hybrid semantic search for code discovery combines embedding generation with keyword extraction to retrieve relevant code chunks. This approach surfaces contextual code patterns and conventions that exact keyword matching alone would miss.

Yes, you can filter semantic code search results by repository, language, and chunk type. These configurable filters refine the hybrid retrieval output to target specific project conventions or SDK implementations.

Find EDNA authentication flows by querying the indexed ASU GitHub organization. The backend surfaces relevant code patterns and integration examples from PeopleSoft, DPL, EEL, Vault, and CI/CD repositories.

Yes, you need pnpm and tsx to run the ASU code discovery backend. These dependencies execute the orchestration scripts that handle embedding generation, backend search, and result caching.

Semantic search for Terraform and CI/CD configurations depends on indexing quality and chunk type filters. Results require validation against actual repository state, as cached embeddings may lag behind recent code commits.