Semantic Vector Search
Efficient querying of embedding collections via semantic similarity search.
Prompt Example
Perform a semantic search in the collection '[COLLECTION-NAME]' (optional) to retrieve information regarding the following concept: [QUERY/CONCEPT].
Concrete examples
Perform a semantic search to identify all documented decisions regarding the Ariadne Academy Library architecture.
This skill enables semantic vector search within specialized embedding collections. It is the primary tool for retrieving information where relevance is determined by semantic meaning rather than exact token matches.
The process relies on comparing query embeddings against indexed vectors. By specifying target collections, you can perform query refinement to narrow the search space, minimize noise, and increase precision.
Best Practices: Use precise, context-rich natural language. Avoid vague queries. Leverage collection filters when the thematic scope is known to optimize both search latency and accuracy.
Technical Notes: This skill performs similarity searches and returns results based on relevance scoring. It is distinct from Long-Term Memory (LTM) lookups, which target explicitly stored, structured facts.