iterative-retrieval

To refine context for subagents, you can use an iterative retrieval pattern that progressively provides detailed codebase context. This approach systematically improves subagent communication and task completion by cycling through context refinement phases up to three times.

The iterative retrieval pattern is a four-phase process that refines codebase context for multi-agent workflows. It systematically improves task completion by providing progressively detailed information to subagents, repeating up to three cycles based on a given task description.

Solving the subagent context problem involves iteratively refining the information passed between subagents. By implementing a four-phase retrieval cycle, workflows can provide progressively detailed codebase context, ensuring subagents have the necessary information to perform effectively.

Iterative context refinement operates through scripts designed to be dispatched within multi-agent workflows. It requires a task description to initiate the four-phase process, making it compatible with workflows that can trigger and pass context to subagents.

The iterative retrieval pattern is limited to a maximum of three repeated cycles for refining context. It specifically addresses providing progressively detailed codebase context for subagents, meaning it requires a clear task description to function effectively.