iterative-retrieval

Name: iterative-retrieval
Availability: InStock
Author: fideguch

Refine codebase context retrieval through a four-phase iterative loop.

1|Updated Jan 30, 2021

Search Tags:#context #codebase #rag #multi-agent #agents #retrieval #iteration

Ready to run

npx skills add https://github.com/fideguch/my_dotfiles --skill iterative-retrieval-fideguch

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: iterative-retrieval
Source: https://github.com/fideguch/my_dotfiles/tree/main/claude/skills/iterative-retrieval
Command: npx skills add https://github.com/fideguch/my_dotfiles --skill iterative-retrieval-fideguch

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Solves the context problem in multi-agent workflows by progressively refining the information subagents need to perform tasks within a codebase.

Core Features & Use Cases

Four-phase retrieval loop (DISPATCH, EVALUATE, REFINE, LOOP) that iteratively identifies relevant files and context.
Cycles with a maximum of three iterations to balance completeness and efficiency.
Applies to codebase exploration, debugging, and feature development where context visibility emerges during task execution.
Supports building retrieval pipelines similar to RAG for code exploration and context-driven agent orchestration.

Quick Start

Start with a broad query and run the 4-phase loop to gather high-relevance files.

Frequently Asked Questions about iterative-retrieval

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

FAQPage Schema

How do I fix missing context when spawning subagents for codebase tasks?▼

Solve the subagent context problem by iteratively retrieving codebase context through a four-phase loop that progressively evaluates and refines file relevance to ensure agents have necessary information.

How does iterative retrieval work for multi-agent codebase exploration?▼

Iterative retrieval works through a four-phase loop: DISPATCH collects candidate files, EVALUATE scores relevance, REFINE adjusts queries, and LOOP repeats up to three cycles to balance completeness with efficiency.

What is the best way to handle context-too-large failures in agent workflows?▼

Handle context-too-large failures by applying iterative retrieval to progressively refine context, scoring file relevance and cycling queries up to three times to extract only high-relevance context for agents.

Can I use iterative retrieval to build RAG pipelines for code exploration?▼

Yes, you can use iterative retrieval to build RAG-like pipelines for code exploration by dispatching broad queries, evaluating file relevance, and refining searches to orchestrate context-driven multi-agent workflows.

What are the limitations of using a three-cycle retrieval loop for subagent context?▼

The limitation of a three-cycle retrieval loop is a maximum of three iterations, which bounds retrieval completeness; if high-relevance files are not found within three cycles, the context may remain incomplete.