verify-prod-after-deploylisted

# Verify Prod After Deploy A deploy is verified when the **changed surface** demonstrably works in prod — not when tests pass, not when the push succeeds, not when health returns 200. ## Why this exists (paid-for lessons) - **T9 insider collector (2026-06-17):** 12 green offline tests, worked on dev, reported live — a prod check found 100% of its SEC fetches 403-ing. A collector that runs but fetches nothing reads as "covered" on every dashboard. - **CI gate (2026-07-08):** Railway deploys are gated on GitHub CI ("Waiting for CI"). A red CI silently blocks every deploy — the push "landing" means nothing. - **Edge outage (2026-07-08):** the app can be healthy internally while the public URL refuses all connections. Health from inside ≠ reachable from outside. ## Steps 1. **CI first.** A push does not deploy until CI is green: `curl -s "https://api.github.com/repos/Murathanx12/Aegis-Finance/actions/runs?per_page=1"` → check `status`/`conclusion` for YOUR head sha. Red CI = no deploy, fix it. 2. **Confirm the commit flipped.** Poll `GET /api/health/full` until `deploy.commit` equals your pushed sha (Railway build ≈ 3-8 min after CI). If it never flips: `railway status` (look for "Waiting for CI" or a stuck deployment). Prefer `aegis_verified_state` (Optimus MCP) when available. 3. **Read the canaries on the same response:** `scheduler.nav.all_fresh` must be true; `recent_warnings` must not contain NEW warnings tied to your change (pre-existing o