How does this help with Claude's context limits?

Instead of sending entire directories, this pattern identifies only the most relevant files (typically scoring 0.7 or higher), ensuring the AI stays within context window limits while having the right information.

What is the Iterative Retrieval pattern?

It is a systematic 4-phase loop (Dispatch, Evaluate, Refine, Loop) that helps AI agents find the most relevant code context by progressively narrowing down search results based on what they learn in each step.

How many cycles does the loop typically run?

The pattern recommends a maximum of 3 cycles. This is usually sufficient to discover relevant terminology and gather enough context without wasting time or resources.

Can this skill handle codebases with unique naming conventions?

Yes, the 'Refine' phase is designed to learn terminology from initial broad searches, allowing the agent to discover and use project-specific terms like 'throttle' instead of 'rate-limit'.

Iterative Context Retrieval

Name: Iterative Context Retrieval
Author: saar120

bysaar120

0•

Herramientas para Desarrolladores

Resolves sub-agent context limitations by progressively refining file searches through a structured multi-phase evaluation loop.

The Iterative Context Retrieval skill implements a robust four-phase workflow—Dispatch, Evaluate, Refine, and Loop—designed to solve the common 'context problem' in multi-agent systems. By starting with broad queries and systematically narrowing down candidates based on relevance scoring and terminology discovery, it ensures Claude has the exact information needed for complex tasks without exceeding context limits. This pattern is particularly effective for navigating large, unfamiliar codebases where initial keyword guesses might miss domain-specific implementation patterns or specific naming conventions.

Características Principales

010 GitHub stars

02Relevance scoring system to prioritize high-impact code context

03Explicit context gap identification to drive subsequent search cycles

04Intelligent exclusion of low-relevance paths to optimize context window

05Four-phase search optimization loop (Dispatch, Evaluate, Refine, Loop)

06Automated terminology discovery for codebase-specific keyword expansion

Casos de Uso

01Efficiently locating bug fix context in large-scale enterprise repositories

02Onboarding AI agents to codebases with unique naming conventions

03Gathering architectural patterns for implementing new cross-layer features

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add saar120/skillgate-registry iterative-retrieval

For use in Claude.ai and ChatGPT

Características Principales

010 GitHub stars

02Relevance scoring system to prioritize high-impact code context

03Explicit context gap identification to drive subsequent search cycles

04Intelligent exclusion of low-relevance paths to optimize context window

05Four-phase search optimization loop (Dispatch, Evaluate, Refine, Loop)

06Automated terminology discovery for codebase-specific keyword expansion

Casos de Uso

01Efficiently locating bug fix context in large-scale enterprise repositories

02Onboarding AI agents to codebases with unique naming conventions

03Gathering architectural patterns for implementing new cross-layer features

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add saar120/skillgate-registry iterative-retrieval

For use in Claude.ai and ChatGPT