iterative-retrieval

Iterative context retrieval solves this by dispatching broad queries, evaluating relevance, and refining search criteria in a loop to gather only the necessary files for subagents. This prevents missing context or context overload errors.

Progressive context refinement dynamically narrows down relevant code snippets through a multi-phase loop. It adapts search queries based on evaluated relevance and identified information gaps up to three times to ensure sufficient context.

It evaluates retrieved content for relevance and identifies information gaps, then dynamically builds and refines search queries. This loop repeats up to three times to ensure subagents receive the precise codebase context required.

Yes, this pattern prevents context too large errors by iteratively narrowing down relevant files and code snippets. It dispatches broad queries and refines search criteria to gather only the exact context needed for the task.

You need progressive context refinement when the exact codebase context required by a subagent is unknown upfront. It dynamically adapts search queries to retrieve precise files without overwhelming the multi-agent workflow with irrelevant data.