tanstack-query

Brains Trust is an AI consultation skill designed to gather second opinions and alternative perspectives from multiple leading AI models. It helps users validate ideas, review technical decisions, challenge assumptions, and identify blind spots across areas such as coding, architecture, debugging, prompting, strategy, and writing. Instead of relying on a single model response, the skill coordinates consultations with models from providers such as OpenAI, Google Gemini, Qwen, DeepSeek, Mistral, and others through OpenRouter or direct provider APIs. The skill is especially useful when users want additional confidence before making important technical or strategic decisions.

The skill supports several consultation patterns depending on the user’s intent. Consensus mode queries two models from different providers in parallel and compares where they agree or disagree. Single mode retrieves one focused opinion from a specified provider or model family, while Devil’s Advocate mode intentionally challenges the current approach or position. The system dynamically selects current leading models by retrieving live model listings from models.flared.au instead of relying on hardcoded model identifiers. It also infers context from the current working session, including recently edited code, architecture discussions, debugging efforts, or prompting workflows, allowing the consultation to adapt automatically to the task at hand.

Brains Trust is intended for developers, technical leads, AI practitioners, prompt engineers, architects, and advanced users who want broader perspective and model diversity during decision-making. Common use cases include peer review of code changes, validating architectural choices, debugging difficult issues, refining prompts, reviewing strategies, or stress-testing assumptions. The skill is particularly valuable in collaborative or high-stakes workflows where identifying weaknesses, trade-offs, or overlooked alternatives can improve overall quality and reliability.