skills.rest
CobaltSato
CobaltSatoCOMMUNITY
Explore all skills by this author →
GitHub

iterative-retrieval

Automate iterative context retrieval with a four-phase loop for subagent workflows.

Updated Jan 30, 2026
Search Tags:#context#codebase#search#subagents#retrieval#iterative
One-click install
npx skills add https://github.com/CobaltSato/avalanche-build-games-tool-kit --skill iterative-retrieval-cobaltsato
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: iterative-retrieval
Source: https://github.com/CobaltSato/avalanche-build-games-tool-kit/tree/main/.claude/skills/iterative-retrieval
Command: npx skills add https://github.com/CobaltSato/avalanche-build-games-tool-kit --skill iterative-retrieval-cobaltsato

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Subagents often launch with limited context and need to discover which files, patterns, and terminology are relevant. The Iterative Retrieval Pattern defines a disciplined loop to progressively refine context by expanding and narrowing search criteria based on evaluated relevance.

Core Features & Use Cases

  • Multi-phase retrieval loop (DISPATCH, EVALUATE, REFINE, LOOP) to identify high-relevance files
  • Dynamic refinement of search terms and focus areas based on evaluated results
  • Robust handling for limited-context subagents in large codebases or datasets

Quick Start

Execute a 3-cycle iterative retrieval: dispatch a broad query, evaluate results for relevance, then refine search criteria and repeat until achieving sufficient high-relevance context.

Frequently Asked Questions about iterative-retrieval

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I discover relevant files in a large codebase when starting with limited context?▼

Iterative context retrieval progressively discovers relevant files by dispatching broad queries, evaluating results, and refining search criteria across multiple cycles to build high-relevance context for subagents.

What is the DISPATCH, EVALUATE, REFINE, LOOP pattern for codebase search?▼

This multi-agent retrieval pattern dispatches a search query, evaluates the relevance of returned files, and refines search terms based on the evaluation, looping through these phases up to three cycles.

How do I refine search criteria dynamically for subagent workflows?▼

You refine search criteria by evaluating initial retrieval results for relevance and dynamically expanding or narrowing focus areas and terminology in subsequent search cycles.

Can I use iterative retrieval to find codebase patterns and terminology across documents?▼

Yes, this approach is designed for limited-context subagents working across large codebases or datasets to dynamically discover relevant files, patterns, and terminology through evaluated retrieval cycles.

What is the maximum number of retrieval cycles allowed when refining context?▼

The iterative retrieval loop runs for a maximum of three cycles to progressively refine search criteria and output high-relevance results for subagent workflows.

Related Skills

Explore More Software Engineering Skills→
wraithyywraithyy

iterative-retrieval

Iteratively refine file retrieval queries across up to three cycles for subagent context.

Community
Intermediate
andrew-starosciakandrew-starosciak
2

iterative-retrieval

Refine subagent context through a four-stage iterative retrieval loop.

Community
Advanced
hbvg234hbvg234
1

subagent-dependency

Coordinate subagent dependencies for parallel execution and safe result merging.

Community
Advanced
gustavofsantosgustavofsantos

ast-grep

Translate natural language queries into ast-grep YAML rules for structural code search.

Community
Intermediate