Can it adapt to unique project naming conventions?

Yes, during the Refine phase, the skill extracts specific keywords and naming patterns found in the initial results to improve the accuracy of subsequent search queries.

What problem does Iterative Retrieval solve?

It solves the 'context problem' where AI agents don't know which files they need until they start working. It prevents context overflow while ensuring the agent doesn't miss critical information.

Why is the loop limited to three cycles?

Three cycles provide the optimal balance between finding deep context and maintaining performance, preventing the agent from getting stuck in an infinite search loop.

How does the relevance scoring work?

Files are scored on a scale from 0.0 to 1.0. High relevance (0.7+) files are kept, while low relevance files are used to refine the search or are explicitly excluded from future cycles.

Iterative Context Retrieval

Name: Iterative Context Retrieval
Author: Fabio29T

byFabio29T

0•

開発者ツール

Resolves sub-agent context limitations by progressively refining file search criteria across multiple evaluation and discovery cycles.

The Iterative Retrieval skill addresses the 'context problem' in multi-agent workflows where agents often lack the necessary project-specific knowledge to perform tasks effectively. By implementing a four-phase loop—Dispatch, Evaluate, Refine, and Loop—this skill enables Claude to move beyond simple keyword searches. It intelligently scores file relevance, identifies information gaps, and learns project-specific terminology on the fly, ensuring that sub-agents receive the most relevant code snippets without exceeding context limits or missing critical dependencies.

主な機能

01Intelligent gap identification to target missing architectural context

020 GitHub stars

03Dynamic keyword and pattern extraction from project-specific files

04Automatic exclusion of irrelevant files to optimize context window

05Four-phase iterative loop: Dispatch, Evaluate, Refine, and Loop

06Automated relevance scoring (0.0 to 1.0) for discovered content

ユースケース

01Gathering implementation patterns and boilerplate for new feature development

02Onboarding sub-agents with precise context for cross-functional architectural changes

03Locating relevant files for complex bug fixes in large, unfamiliar codebases

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add fabio29t/everything-claude iterative-retrieval

For use in Claude.ai and ChatGPT

主な機能

01Intelligent gap identification to target missing architectural context

020 GitHub stars

03Dynamic keyword and pattern extraction from project-specific files

04Automatic exclusion of irrelevant files to optimize context window

05Four-phase iterative loop: Dispatch, Evaluate, Refine, and Loop

06Automated relevance scoring (0.0 to 1.0) for discovered content

ユースケース

01Gathering implementation patterns and boilerplate for new feature development

02Onboarding sub-agents with precise context for cross-functional architectural changes

03Locating relevant files for complex bug fixes in large, unfamiliar codebases

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add fabio29t/everything-claude iterative-retrieval

For use in Claude.ai and ChatGPT