What problem does it solve? When supercodex delegations fail, users cannot tell whether the cause is a missing Codex CLI, an outdated version, a broken login, or unavailable GPT-5.6 models. This Skill runs a structured health check and reports exactly which component is broken and how to fix it. ## Core Features & Use Cases - Environment health check: Runs the codex-companion script in setup mode and interprets JSON fields covering Node.js availability, Codex CLI installation, CLI version versus the GPT-5.6 minimum, and login status. - Live model probe: Dispatches a real GPT-5.6 task expecting a PONG response to confirm end-to-end delegation works, catching model-rejection errors that version checks miss. - Actionable reporting: Summarizes component status in a table with tailored fix commands, and declares one of three outcomes: fully ready, fallback mode on Claude subagents, or running on a non-GPT-5.6 default model. - Use Case: After installing supercodex, ask Claude to run the setup check; it reports that your Codex CLI is below the minimum version and gives you the exact npm upgrade command. ## Quick Start Ask Claude to run the supercodex setup check to verify the Codex CLI, authentication, and GPT-5.6 model availability on this machine.