How does the brainstorming skill handle model failures?

It utilizes a resilient fallback chain that automatically attempts alternative models like Gemini, Grok, or Deepseek if the primary provider fails to respond.

What is the Consensus Matrix feature?

It is a mathematical algorithm that clusters diverse AI suggestions, counts agreement across different models, and calculates a weighted confidence score to identify the strongest architectural path.

Do I need an OpenRouter account to use this skill?

Yes, an OpenRouter API key is recommended to access the full range of explorer models and ensure the multi-model validation works correctly.

Can I customize the confidence thresholds for automation?

Absolutely. You can adjust parameters like 'confidence_threshold_auto' to define when the skill can proceed automatically and when it must stop for user approval.

Does this skill work with git worktrees?

Yes, it is designed to work alongside git-worktree skills to allow for parallel exploration and planning without disrupting your main development branch.

Resilient Brainstorming & Planning

Name: Resilient Brainstorming & Planning
Author: MadAppGang

byMadAppGang

•

生産性とワークフロー

Automates high-reliability architectural planning and collaborative ideation using multi-model consensus and fallback chains.

This skill facilitates robust technical planning and architectural design by leveraging multiple AI models simultaneously. It overcomes reliability gaps in standard AI prompting by using a resilient fallback chain including Claude, Gemini, Grok, and Deepseek to ensure completion during API outages. Through a structured five-phase workflow—including problem analysis, parallel exploration, and mathematical consensus scoring—it validates technical ideas against multiple perspectives to provide high-confidence implementation roadmaps.

主な機能

01Adaptive confidence-based gating to automatically proceed or request user intervention

02Parallel exploration patterns to generate diverse architectural solutions 3-5x faster

033 GitHub stars

04Mathematical consensus matrix and confidence scoring for objective approach validation

05Multi-model fallback chains (Claude, Gemini, Grok, Deepseek) for 100% planning uptime

06Task-driven structured interaction that eliminates fragile UI dependencies

ユースケース

01Evaluating multiple implementation strategies for performance and scalability trade-offs

02Creating detailed, section-by-section implementation plans with validated success criteria

03Designing complex system architectures or microservices from initial requirements

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add madappgang/magus planning

For use in Claude.ai and ChatGPT