See burn. Act before warm storage dies.
Operate the loop: Pay runway → alert fire → policy dry-run → projected cold promotions. Demo model couples FinOps to tier so promotions show runway regained.
Runway
Coupling · promote → runway
Activity feed
CLI mirror
$ foc-ops status — waiting…
FinOps + alert drill
Same gauges as Prometheus. Inject spend to watch ok → low_runway → deficit fire in-console.
Funds vs lockup
Live /metrics scrape
scraping…
PromQL that fires: foc_ops_alert > 0 or
foc_ops_runway_days < 14
Tier dry-run · warm → cold
Scrub policy — rows migrate live. Then Run dry-run for stepped intents (no chain write).
Promote vs skip
Age vs threshold
Projected registry intents
Would promote
| Piece | Age | Size | Reason |
|---|
Skipped
| Piece | Reason | Tier |
|---|
Network + FVM registry
Always-on Calibration RPC probe (Glif). TierRegistry deploy + wallet recordPromotion when address is set.
Live RPC probe
TierRegistry
Deploy with scripts/deploy-registry.mjs when you have Calibration tFIL.
Wallet tx template
connect wallet + deploy registry to enable on-chain promote
What you just ran
Live Pay + dry-run work now. TierRegistry is on Calibration; Mainnet + warm→cold execute ship in the grant.
Read Pay health
Live Calibration getAccountSummary (or labeled drills).
Alert thresholds
Inject spend until badges flip; metrics pane mirrors Prometheus.
Tier dry-run
Policy scrub + Run dry-run produces promote intents.
FVM registry
Calibration TierRegistry live — wallet can recordPromotion.