Overview
Semantic Search lets you find code in your project by describing what you’re looking for in plain English. Instead of exact string matching, it understands the meaning of your query. Use Cmd+P to activate.Semantic Search goes beyond
grep. Searching for “authentication flow” finds login logic, token refresh handlers, and session management — even if those words don’t appear in the code.How It Works
Parsaa indexes your codebase
When you enable Semantic Search for a project, Parsaa generates vector embeddings for your code using VoyageAI (or OpenAI embeddings). This initial indexing runs once.
Embeddings are stored locally
All embeddings are saved on your Mac at
~/Library/Application Support/Parsaa/Embeddings/. They never leave your machine.You search in natural language
Press Cmd+P and type what you’re looking for in plain English. Your query is converted to an embedding.
Setup
Semantic Search requires a VoyageAI API key.Get a VoyageAI API key
Sign up at voyageai.com and generate an API key.
Configure in Parsaa
Open Parsaa’s settings and add your VoyageAI API key in the embeddings configuration section.
You can also use OpenAI embeddings as an alternative to VoyageAI. Configure this in Parsaa’s settings.
Example Queries
Architecture & Flow
Architecture & Flow
- “authentication flow”
- “how data gets saved to Core Data”
- “navigation between screens”
- “dependency injection setup”
Specific Patterns
Specific Patterns
- “network error handling”
- “SwiftUI views that display user data”
- “async/await usage in the networking layer”
- “custom view modifiers”
Business Logic
Business Logic
- “Core Data save logic”
- “payment processing”
- “user session management”
- “push notification handling”
Privacy
Local Storage
Embeddings are stored locally on your Mac and never uploaded to Parsaa’s servers.
Zero Data Retention
Code sent to VoyageAI for embedding generation is processed and immediately discarded — VoyageAI maintains zero data retention for embedding requests.
