Indexing
Wingman includes a vector database built in that allows you to configure inclusion and exclusion filters for files to be embedded using the embedding provider of your choice. Once you configure your embeddings provider, you can choose to run a full index of your codebase from the Indexer view, or Wingman will index your files as your save them.
Files will not be re-indexed on save if their contents have not changed.
Wingman will use cheaper models for indexing code files, for Anthropic it will use Claude-3.5-Haiku, and for OpenAI it will use gpt-4o-mini
Wingman is the only code assistant that not only embeds files, but creates a comprehensive graph of your entire codebase. Allowing Wingman to leverage file to file relationships.
Embeddings are stored locally.