ship-gatelisted

# Agent production readiness review A demo that works in a keynote is not an agent that works at 2am. This skill runs the review that tells the difference. It scores an agent across seven dimensions and returns the gaps that stand between a promising prototype and a dependable production system. It is the ship-time gate. Its sibling, the [agent-or-workflow](https://github.com/hamidettefagh/agent-or-workflow) skill, is the design-time gate. This one decides whether an agent is ready to run; that one decides what to build in the first place. ## When to run this Run this when the user is: - Preparing an agent for production or a go-live. - Asking whether an agent is ready, or good enough to ship. - Reviewing an agent's design, prompt, tools, or deployment. - Turning a working prototype into something they can operate. Do not run it against a throwaway prototype the user has said will never reach production. Ask if you are unsure. ## How to run the review Work through four phases in order. Do not skip to a score. ### Phase 1: Understand the agent Before assessing anything, establish the basics, either by asking the user or by reading the material they provide: - What the agent does, and for whom. - Where it runs, and what it is connected to: tools, data, systems of record. - What actions it can take, and which of those are hard to reverse. - What "wrong" looks like, and who is harmed when it happens. Keep this short. You are calibrating how strict the rest of the revi