ecc-iterative-retrieval

To discover relevant files in a multi-agent workflow with unknown context, apply an iterative retrieval pattern that progressively refines subagent context through a four-phase loop of DISPATCH, EVALUATE, REFINE, and LOOP.

The best way to narrow down relevant codebase patterns across multiple agents is using iterative retrieval, which learns terminology and patterns from initial searches to guide subsequent context discovery and relevance-based selection.

The four-phase iterative retrieval loop refines context by dispatching search tasks, evaluating the relevance of returned information, refining the search parameters based on findings, and looping until cycle limits are met.

You need iterative context discovery for multi-agent systems when subagents start with limited context and must progressively reveal information to identify relevant files, patterns, and terminology within an unknown codebase.

Yes, iterative retrieval enforces strict cycle limits on context refinement cycles, ensuring the progressive discovery of relevant files terminates efficiently while applying relevance-based selection to filter results.

Yes, you can use this context refinement pattern to integrate a new task across multiple agents by efficiently narrowing down the most relevant files and concepts through incremental context discovery and evaluation.