How does iterative retrieval improve Claude's performance?

Instead of guessing what files are needed, Claude performs a broad initial search, evaluates the results, and uses discovered keywords to refine its search until it has high-relevance context.

Can I define my own relevance criteria?

Yes, the skill utilizes a scoring system (High, Medium, Low) that can be adjusted via agent prompts to prioritize specific types of files or patterns.

How many cycles does the retrieval process take?

The skill is optimized to run for a maximum of three cycles, which is typically sufficient to find high-relevance context while maintaining execution speed.

What is the 'context problem' in AI agent workflows?

The context problem occurs when subagents are spawned without knowing which files are relevant to their task, leading to either missing information or context window overflow if too much data is provided.

Does this skill work with any programming language?

Yes, the Iterative Retrieval pattern is language-agnostic and relies on file patterns and keyword discovery applicable to any codebase.

Iterative Context Retrieval

Name: Iterative Context Retrieval
Author: affaan-m

byaffaan-m

•

43,117

生産性とワークフロー

Refines AI context retrieval through an intelligent four-phase feedback loop to solve the context problem in complex multi-agent workflows.

概要

The Iterative Retrieval skill addresses the critical 'context problem' encountered in multi-agent workflows where subagents often lack the specific codebase knowledge needed to perform their tasks. By implementing a standardized Dispatch-Evaluate-Refine-Loop cycle, this skill allows Claude to progressively discover relevant files, learn project-specific terminology, and identify architectural patterns without exceeding context limits. It ensures that subagents operate with high-precision information, making them significantly more effective at resolving bugs and implementing features in large, unfamiliar repositories.

主な機能

Dynamic terminology discovery to adapt to project-specific naming conventions
Four-phase autonomous retrieval cycle: Dispatch, Evaluate, Refine, and Loop
Intelligent relevance scoring (0-1.0) to prioritize high-value code files
Automated gap analysis to identify and target missing architectural context
43,117 GitHub stars
Optimized three-cycle execution limit to balance precision with performance

ユースケース

Onboarding subagents to specific architectural patterns and internal APIs
Navigating large-scale repositories where relevant file locations are unknown
Investigating complex bugs that span multiple loosely-coupled modules

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add affaan-m/everything-claude-code iterative-retrieval

For use in Claude.ai and ChatGPT

Download Skill

GitHub

概要