Runs the final ship-or-hold gate on any creative artifact - performance budgets, accessibility floors, smallest-surface passes, the honesty contract on screenshots and claims, and a sweep of the named anti-pattern tells - and returns a ship or hold verdict with every blocking item listed and routed. Use when someone asks "is this creative work ready to ship", "run the final checks on our launch page", "what do we verify before the campaign goes live", "gate these assets before we publish", or in the last review pass before any launch or publish moment. Do NOT use for launch-day operations, timing, and rollbacks - use launch-day-runbook instead; for verifying a build against its design spec, use design-qa-checklist; for the quality judgment itself, use creative-direction-review; for brand-rule compliance on one artifact, use brand-application-review; for drift across a multi-asset campaign, use campaign-coherence-review.
---
name: pre-ship-creative-gate
description: Runs the final ship-or-hold gate on any creative artifact - performance budgets, accessibility floors, smallest-surface passes, the honesty contract on screenshots and claims, and a sweep of the named anti-pattern tells - and returns a ship or hold verdict with every blocking item listed and routed. Use when someone asks "is this creative work ready to ship", "run the final checks on our launch page", "what do we verify before the campaign goes live", "gate these assets before we publish", or in the last review pass before any launch or publish moment. Do NOT use for launch-day operations, timing, and rollbacks - use launch-day-runbook instead; for verifying a build against its design spec, use design-qa-checklist; for the quality judgment itself, use creative-direction-review; for brand-rule compliance on one artifact, use brand-application-review; for drift across a multi-asset campaign, use campaign-coherence-review.
---
# pre-ship-creative-gate
The most public failures are the cheapest to catch: a hero that takes 4 seconds to paint, a screenshot with impossible data that one journalist screenshots back, a favicon that vanishes at 16px, a launch page with no reduced-motion path. Reviews judge whether work is good; the gate verifies it is safe to publish. Work that passed every review still fails launches on exactly these mechanical items, because nobody owned the last pass.
## Operating procedure
Run the steps in this order because it is fix-time-descending: performance regressions take days to fix, copy receipts take minutes, and running the slow fixes first protects the ship date. Every check below declares its evidence mode - COMPUTED by the embedded script from supplied values, JUDGED yes/no from a supplied artifact, or ELICITED as a supplied measurement with a named method. A check with no evidence is not a pass; it is an unverified hold.
### Step 1: Gather the charter, the artifacts, and the measurements
Ask for each of these before checking anything; if the user cannot supply an item, record it as missing rather than working around it.
1. **Charter** - paste the charter block if one exists. If none exists, the universal floors below still bind in full; route the client to premium-design-foundations to establish a charter, and mark every charter-conditional taste check N/A rather than substituting house taste.
2. **Artifacts** and their publish surfaces - the page, stills, deck, or film actually going live, not the mockups.
3. **Ship moment** - the gate runs at least 48 hours before it, so a hold has room to become a fix, not a slip (worked example - Meridian ships Tue 12:01 AM PT).
4. **Measured performance numbers** if a live URL exists - a Lighthouse or WebPageTest run on a mid-tier device, file sizes, a profiler trace. See references/measurement-methods.
5. **Launch precondition numbers** from the launch plan, verified against their own thresholds - a waitlist count decides channel strategy (VELOCITY-PLAY vs FEEDBACK-PLAY per waitlist-and-teaser-mechanics), never the ship date, so the gate checks that the number is real, not that it clears some floor.
6. **Which reviews already ran** - the gate assumes creative-direction-review passed; it does not re-judge merit.
If a number is a guess, label it a guess - guesses hold the gate, they do not pass it.