How does the 3-stage voting pipeline work?

First, all models generate answers independently; second, they peer-rank each other's responses; finally, a chairman model synthesizes the data into one optimal decision.

Is the agent-council CLI required to use this skill?

Yes, you must have the agent-council package installed globally via npm for the skill to execute the decision pipeline.

Do I need API keys for all models?

Yes, the agent-council CLI requires access to the respective APIs for Claude, Codex, and Gemini to function correctly within your environment.

What is the benefit of using multiple models for a decision?

Using multiple models like Claude, Codex, and Gemini reduces the risk of hallucinations and individual model bias, providing a more balanced and reliable technical recommendation.

When should I invoke the council-decision skill?

Use it for high-stakes decisions like architectural patterns, database selection, or evaluating trade-offs between complex implementation strategies.

Agent Council Decision

Name: Agent Council Decision
Author: mylukin

bymylukin

•

Productivity & Workflow

Facilitates consensus-driven decision-making by orchestrating multiple AI models to evaluate complex architectural and implementation questions.

About

The Agent Council Decision skill integrates the agent-council CLI into the Claude environment, enabling a structured three-stage voting pipeline to resolve difficult technical dilemmas. By simultaneously querying Claude, Codex, and Gemini, the skill subjects each independent response to peer ranking before a chairman model synthesizes a final recommendation. This approach effectively minimizes single-model bias and provides developers with a robust, multi-perspective framework for making critical choices regarding system architecture, technology stacks, and complex design patterns.

Key Features

Multi-model orchestration using Claude, Codex, and Gemini
Three-stage consensus pipeline including independent response and peer-ranking
Automated synthesis of findings into a single, authoritative recommendation
Bias reduction by evaluating diverse AI perspectives simultaneously
Direct CLI integration for seamless use during the development lifecycle
20 GitHub stars

Use Cases

Evaluating architectural design trade-offs between different system structures
Selecting the optimal technology stack or database for a new project
Comparing multiple implementation approaches for complex feature sets

About

Key Features

Multi-model orchestration using Claude, Codex, and Gemini
Three-stage consensus pipeline including independent response and peer-ranking
Automated synthesis of findings into a single, authoritative recommendation
Bias reduction by evaluating diverse AI perspectives simultaneously
Direct CLI integration for seamless use during the development lifecycle
20 GitHub stars

Use Cases

Evaluating architectural design trade-offs between different system structures
Selecting the optimal technology stack or database for a new project
Comparing multiple implementation approaches for complex feature sets