iterative-retrieval

To narrow down context for subagents, you can use iterative search cycles that score, filter, and select relevant files, ensuring focused retrieval without overwhelming the system with entire data trees.

The best way to refine search results is by performing up to three iterative search cycles, ranking files based on relevance to your query to prioritize key information for precise context gathering.

You perform iterative context retrieval by initiating a search focused on specific project files, allowing the system to score and refine the selection over multiple cycles to gather precise context for new feature development.

Yes, you can use iterative retrieval to improve decision-making when managing large documents by refining search results iteratively to extract focused and relevant context snippets for better summarization.

A limitation of multicycle search for context selection is that the iterative refinement process is capped at three search cycles, which may restrict depth when querying extremely large or complex datasets.