What problem does it solve? Subagents in multi-agent workflows are spawned without knowing which files, patterns, or terminology they need, so sending everything blows context limits while sending nothing leaves them blind. This Skill provides a structured loop that progressively discovers the right context. ## Core Features & Use Cases - Four-Phase Retrieval Loop: Dispatch broad queries, evaluate file relevance on a 0-1 scale, refine search criteria from discovered terminology, and loop up to 3 cycles. - Relevance Scoring & Gap Tracking: Scores files as high/medium/low relevance and explicitly identifies missing context to drive the next refinement cycle. - Terminology Discovery: Learns codebase-specific naming conventions (e.g., "throttle" instead of "rate limit") during early cycles to improve later searches. - Use Case: When fixing an authentication token expiry bug, start with broad "auth/token" searches, discover jwt-utils and session-manager files through refinement, and stop once enough high-relevance files are gathered. ## Quick Start Use the iterative-retrieval pattern to find the right codebase context for this subagent task, starting broad and refining over up to three cycles.
