Create a Semantic Search Tool from Documents — create_search

Convenience function that creates a ragnar store from documents and wraps it in a search tool in one step.

Usage

create_search_tool(
  documents,
  embedding_fn,
  k = 5L,
  name = "search_documents",
  description = NULL,
  ...
)

Arguments

documents: Character vector of documents, or a data frame with a 'text' or 'content' column.
embedding_fn: Embedding function from ragnar (e.g., ragnar::embed_openai()).
k: Number of documents to retrieve per search (default 5).
name: Tool name (default "search_documents").
description: Optional tool description.
...: Additional arguments passed to ragnar::ragnar_store_create().

Value

A search tool function.

Examples

if (FALSE) { # \dontrun{
# Create tool directly from documents
docs <- c(
  "R is a programming language for statistical computing.",
  "Python is a general-purpose programming language.",
  "Julia is designed for high-performance numerical computing."
)

search_tool <- create_search_tool(
  documents = docs,
  embedding_fn = ragnar::embed_openai(),
  k = 2
)

# Use the tool
search_tool("What language is best for statistics?")
} # }