Semantic Search

During indexing, QEEK builds searchable representations (embeddings) of code and related metadata. When you ask a question in Chat or generate docs, retrieval finds passages that match the meaning of your query, not only exact keywords.

Practical search often combines semantic similarity with keyword/structural signals (and AST-derived relationships). That mix is why answers can cite specific files even when you describe a problem in product language.

Embeddings and analysis support search and documentation. QEEK does not use your code to train foundation models. Details on retention and encryption are on the Security and Privacy pages.