iterative-retrieval

Progressive context retrieval solves subagent information gaps by iteratively dispatching broad queries, evaluating candidate files for relevance, and refining search criteria based on identified terminology. This multi-agent system pattern loops up to three times to gather sufficient context for complex tasks.

Iterative search for codebase navigation dynamically refines context retrieval by dispatching broad queries, evaluating candidate files, and looping up to three times. This pattern progressively identifies relevant code, configuration, and documentation files to solve the subagent context problem.

Refine search criteria by dispatching broad queries first, then evaluating candidate files for relevance, and identifying gaps in terminology. The iterative retrieval pattern loops up to three times to intelligently gather all relevant code and documentation for focused debugging context.

Iterative context retrieval works for user authentication and session management by progressively searching and evaluating files to build relevant context. It dispatches broad queries, refines search criteria based on identified gaps, and loops to gather all related configuration and code files.

Iterative retrieval for agent prompting limits loops to three iterations to balance context gathering efficiency. While it dynamically refines search criteria and evaluates candidate files, complex workflows may still require additional manual context evaluation if the subagent context problem persists beyond the loop limit.