Do I need separate API keys for each model?

Yes, to use the full Council Mode, you must provide your own API keys for OpenAI, Google Gemini, Perplexity, and xAI in your environment configuration.

What is the LLM Council pattern used by this skill?

It is an architecture where multiple frontier AI models analyze a single query in parallel, review each other's work for accuracy, and have a final 'chairman' model synthesize the results into a single comprehensive report.

How long does a research task typically take?

Simple lookups using Perplexity take about 15 seconds, while a full Council Mode orchestration with peer review and synthesis typically completes in 60 to 90 seconds.

Which AI models are supported in the research council?

The skill currently orchestrates GPT-5.1, Gemini 3.0 Pro, Perplexity Sonar, and Grok 4.1, alongside Claude's native reasoning capabilities.

Can I save the research results to my local files?

Yes, the skill is designed to automatically format results into Markdown and save them to a specified Obsidian vault using the Obsidian MCP tool.

Multi-Model Research Council

Name: Multi-Model Research Council
Author: krishagel

bykrishagel

生産性とワークフロー

Orchestrates a council of frontier LLMs to conduct deep-dive research with cross-model validation and automated synthesis.

概要

The Multi-Model Research Council skill implements Karpathy's LLM Council pattern within Claude Code to deliver superior research outcomes. By leveraging the unique strengths of multiple frontier models—including GPT-5.1, Gemini 3.0 Pro, Perplexity Sonar, and Grok 4.1—it conducts parallel queries, facilitates model-to-model peer reviews, and performs a final synthesis. This approach is ideal for complex technical analysis, factual verification, and strategic research where a single model's perspective might be insufficient or biased.

主な機能

Adaptive routing between quick factual lookups and deep council analysis
Chairman-style synthesis for a single, cohesive, and balanced report
0 GitHub stars
Seamless integration with Obsidian for structured knowledge management
Automated peer review system where models evaluate and rank each other's findings
Parallel orchestration of GPT-5.1, Gemini 3.0, Perplexity, and Grok

ユースケース

Comparative analysis of industry strategies or competing technology stacks
Generating comprehensive research reports with citations directly into Obsidian
High-stakes technical research requiring cross-model factual verification

概要

主な機能

Adaptive routing between quick factual lookups and deep council analysis
Chairman-style synthesis for a single, cohesive, and balanced report
0 GitHub stars
Seamless integration with Obsidian for structured knowledge management
Automated peer review system where models evaluate and rank each other's findings
Parallel orchestration of GPT-5.1, Gemini 3.0, Perplexity, and Grok

ユースケース

Comparative analysis of industry strategies or competing technology stacks
Generating comprehensive research reports with citations directly into Obsidian
High-stakes technical research requiring cross-model factual verification