When is it best to use recursive reasoning?

Use it for multi-step problems that benefit from planning, execution, and verification, such as complex refactoring or building new modules from scratch.

How does the Master/Sub-Agent architecture work?

The Master agent acts as a planner and verifier, while Sub-Agents (recursive-executors) handle the actual implementation of specific steps defined in the master plan.

What is the primary benefit of the Recursive Reasoning skill?

It allows Claude to handle extremely complex tasks by breaking them into smaller, verifiable sub-tasks, preventing errors that often occur in single-pass AI generations.

Can this skill fix its own mistakes?

Yes, it uses a Self-Refine loop where the Master agent critiques sub-agent output and re-delegates the task if the quality or correctness standards are not met.

Which tools does this skill utilize?

The skill leverages the Read tool, the Task tool for delegation, and sequential thinking for structured analysis and planning.

Recursive Reasoning Orchestrator

Name: Recursive Reasoning Orchestrator
Author: lollipopkit

bylollipopkit

•

生産性とワークフロー

Decomposes complex, multi-step problems into manageable sub-tasks delegated to specialized agents for execution and verification.

The Recursive Reasoning skill implements a sophisticated Master/Sub-Agent architecture that transforms Claude into a central orchestrator for high-complexity engineering tasks. By utilizing advanced reasoning techniques such as Least-to-Most prompting and Reflexion, the skill acts as a 'Brain' that plans, delegates to recursive executors, and rigorously verifies every output before synthesis. This approach is ideal for large-scale refactoring, multi-file feature development, or any scenario where a single-pass solution might fail due to task density or complexity.

主な機能

01Automated task decomposition using Least-to-Most reasoning

02Integration with sequential thinking for deep logical planning

03Master/Sub-Agent orchestration for structured problem solving

045 GitHub stars

05Reflexive memory buffer to track performance and insights

06Iterative Self-Refine loops for quality control and verification

ユースケース

01Deep research and synthesis tasks across vast directory structures

02Large-scale codebase migrations requiring step-by-step verification

03Complex feature implementation involving multiple dependencies and tests

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add lollipopkit/cc-plugins recursive

For use in Claude.ai and ChatGPT

Download Skill