Can this skill run autonomously without human input?

No, the current version specifically removes autonomous modes (like the previous Sisyphus Protocol) to ensure all high-stakes research decisions remain under direct human control.

How many specialized agents are included?

The orchestrator manages 24 specialized agents across 8 categories, ranging from Theoretical Framework Architects to Paper Retrieval and Screening Assistants.

What is a Research Orchestrator checkpoint?

A checkpoint is a mandatory pause in the workflow where the AI stops execution to wait for explicit human approval on critical decisions like methodology, theory selection, or research direction.

Does it support PRISMA for systematic reviews?

Yes, it includes Category I agents specifically designed to handle the review pipeline, including database selection, screening criteria, and RAG-based paper analysis.

How does it optimize for cost and performance?

The orchestrator uses tiered model routing, assigning complex reasoning to Claude Opus, standard tasks to Sonnet, and simple data extraction to Haiku.

Research Orchestrator

Name: Research Orchestrator
Author: HosungYou

byHosungYou

0•

Ciencia de Datos y ML

Manages complex academic research workflows through 24 specialized agents with mandatory human-in-the-loop checkpoints.

Research Orchestrator is a sophisticated framework for Claude Code designed to handle end-to-end academic research, from refining research questions to systematic reviews and meta-analyses. It utilizes a human-centered approach, ensuring that critical decisions—such as paradigm selection or methodology approval—require explicit human confirmation through gated checkpoints. By routing tasks across 24 specialized agents optimized for Opus, Sonnet, and Haiku models, it balances deep theoretical reasoning with efficient data extraction and academic style auditing, ensuring high-quality, rigorous research outputs.

Características Principales

01Mandatory human-in-the-loop checkpoints for critical research decisions

02Specialized 24-agent library covering foundation, evidence, design, and analysis

03Automated Systematic Review (PRISMA) and Meta-Analysis pipelines

04Tiered model routing (Opus/Sonnet/Haiku) for cost and performance optimization

05Academic style auditing and humanization pipeline for publication readiness

060 GitHub stars

Casos de Uso

01Automating the academic publication pipeline including journal matching and style auditing

02Conducting comprehensive systematic literature reviews and meta-analyses with PRISMA compliance

03Designing and refining quantitative, qualitative, or mixed-methods research frameworks

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add hosungyou/diverga research-orchestrator

For use in Claude.ai and ChatGPT

Download Skill