What models does this skill support?

The skill is designed to optimize research across a multi-model ecosystem, specifically assigning tasks to Claude (synthesis), Gemini (grounding/breadth), and GPT (technical depth).

What is the WWHTBT protocol?

WWHTBT stands for 'What Would Have To Be True.' It is an analytical framework used by the skill to resolve conflicting claims between different AI models or data sources.

Does this skill provide automated web searching?

This skill focuses on the design and consolidation phases of research. It creates the prompts and strategies used for searching and then synthesizes the results once they are gathered.

Can I use this for technical architecture research?

Yes, it includes a specific 'Technology Evaluation' MECE pattern that covers capability, maturity, fit, cost, and risk/governance.

Multi-LLM Research Brief & Synthesis

Name: Multi-LLM Research Brief & Synthesis
Author: Agentient

byAgentient

0•

생산성 및 워크플로

Designs optimized multi-model research strategies and synthesizes complex findings into actionable intelligence reports.

The Research Brief skill automates a sophisticated two-phase research lifecycle within Claude Code, designed for high-stakes strategic and technical analysis. In Phase 1, it decomposes complex objectives into MECE (Mutually Exclusive, Collectively Exhaustive) structures and assigns specific components to the most capable LLMs—leveraging Claude for nuance, Gemini for breadth, and GPT for depth. In Phase 2, it ingests these multi-model outputs to resolve conflicting data using the 'What Would Have To Be True' (WWHTBT) protocol, scores evidence strength, and identifies 'unknown unknowns,' ultimately producing decision-ready reports with tiered confidence levels.

주요 기능

01MECE research decomposition for market, competitive, and technical analysis

02Automated multi-model task routing based on specific LLM performance strengths

03WWHTBT (What Would Have To Be True) conflict resolution for disagreeing models

040 GitHub stars

05Evidence Strength Tribunal for scoring source quality on a 5-point rubric

06Comprehensive gap analysis and 'unknown unknowns' discovery probes

사용 사례

01Executing multi-model competitive intelligence to map product roadmaps and GTM strategies

02Conducting strategic market research with built-in risk assessment and bias auditing

03Synthesizing conflicting technical research into a unified technology evaluation

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add agentient/vibekit create-research-brief

For use in Claude.ai and ChatGPT

Download Skill