How is file relevance determined?

Files are scored on a scale of 0 to 1 based on how directly they implement target functionality or contain related patterns and types.

How many refinement cycles does the skill use?

The pattern is designed to run for a maximum of 3 cycles, which is typically sufficient to find the necessary files without significantly increasing latency.

Can this skill adapt to projects with non-standard terminology?

Yes, the Refine phase specifically extracts keywords and patterns from the first set of results to improve subsequent search queries.

What problem does Iterative Retrieval solve?

It solves the context problem where agents either have too little information to work effectively or too much irrelevant information, which wastes tokens and leads to errors.

Iterative Context Retrieval

Name: Iterative Context Retrieval
Author: nbossn

bynbossn

0•

Developer Tools

Implements a multi-phase refinement loop to provide AI agents with precise codebase context while minimizing token waste.

The Iterative Retrieval skill solves the 'context problem' in multi-agent workflows where subagents often lack the necessary information to begin work or are overwhelmed by irrelevant data. By employing a four-phase loop—Dispatch, Evaluate, Refine, and Loop—this pattern allows Claude to progressively discover relevant files, learn project-specific terminology, and identify gaps in its own knowledge. This systematic approach ensures that agents receive high-relevance context (scoring 0.7 or higher) for complex tasks like bug fixing or feature implementation without exceeding token limits or incurring unnecessary costs.

Key Features

01Four-phase refinement loop for progressive context discovery

020 GitHub stars

03Explicit gap identification to resolve missing context failures

04Automatic terminology discovery to adapt to project-specific naming

05Token usage optimization through targeted file selection

06Quantitative relevance scoring system to prioritize critical code

Use Cases

01Orchestrating subagents that require precise, minimal context to perform tasks

02Implementing new features in large codebases with unfamiliar patterns

03Fixing complex bugs that span multiple unknown modules or files

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nbossn/claude-code-base iterative-retrieval

For use in Claude.ai and ChatGPT

Key Features

01Four-phase refinement loop for progressive context discovery

020 GitHub stars

03Explicit gap identification to resolve missing context failures

04Automatic terminology discovery to adapt to project-specific naming

05Token usage optimization through targeted file selection

06Quantitative relevance scoring system to prioritize critical code

Use Cases

01Orchestrating subagents that require precise, minimal context to perform tasks

02Implementing new features in large codebases with unfamiliar patterns

03Fixing complex bugs that span multiple unknown modules or files

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add nbossn/claude-code-base iterative-retrieval

For use in Claude.ai and ChatGPT