adeilsonrbrito/cross-model-consult

Cross Model Consult is a set of skills for the Claude Code AI coding assistant that let it ask other AI coding tools, specifically Codex from OpenAI, Grok from xAI, and OpenCode, for a second opinion on hard decisions. The idea behind the project is that asking the same AI model to double check its own answer is not real independent verification, since it reasons using the same training and the same blind spots. Getting an answer from a genuinely different model, asked cold with no shared context, is much stronger evidence, especially when models agree or clearly disagree with each other. The project includes individual skills for consulting each of the three outside tools on its own, plus a panel skill that sends the same question to all three in parallel, without letting them see each other's answers, and then summarizes where they agree and where they differ. There is also a baseline skill that gives Claude Code permission to decide on its own when a decision is expensive or hard to reverse enough to justify calling in outside opinions, while skipping consults for routine or trivial work. Using any of these tools requires you to separately install and log into each underlying command line tool yourself: the Codex CLI needs an OpenAI account, Grok CLI needs an xAI account, and OpenCode needs whichever provider account you configure it with. You do not need all three installed, since each skill works independently and the panel can run with just a subset of them. Installation is done by cloning the repository and creating symbolic links from the skill folders into your Claude Code skills directory. Every wrapper script defaults to read only access, since a consult is meant to give advice, not take action, and each skill's model choice can be changed through environment variables or by editing the script directly. The project is open source under the MIT license.