# A green deploy is not proof the checks ran

Status: DRAFT, nothing posted. Single-image infographic, 2 slide(s), no cover. Idea IB-0284.

## What it shows

Slide 1 shows the deploy gate verdict counts from the 9 October review (14 September onward): most deploys passed because the gate could not read the target, not because every check ran. Slide 2 is how to find out.

No numbers appear on the slides unless listed below.

## Claims and sources

| Claim | Source (agency-operator unless stated) | Read-date |
|---|---|---|
| cf-deploy-gate: 71 fail-open vs 29 pass vs 18 block since 14 September (as read 2026-10-09; the fix was proposed as item A2) | `docs/instructions-review-2026-10-09.md` (A2) | 2026-10-11 |
| A green deploy is still not proof the checks ran: hooks.tsv names what each gate did | `.claude/rules/14-cf-pages-economy.md` (fail posture) | 2026-10-11 |
| Gates fail open when they cannot resolve the deploy target | `.claude/rules/14-cf-pages-economy.md` | 2026-10-11 |