Ideas for additional yvUSD alerting

1. Leveraged looper liquidation alerting: if a strategy is 4x leveraged, we should have monitoring to prevent a liquidation spiral if we get too near to liquidation
2. Looping strategy parameter changes: it might be good to alert when the leverage amount is changed in a looping strategy, for example if targetLeverageRatio or maxLeverageRatio moves from 5x to 10x leverage
3. Detecting cross-chain losses: the health check in BaseHealthCheck.sol:165 uses a lossLimit of 0%, so a harvest with a loss will revert. We should alert on this scenario (maybe we already alert on a reverting harvest in another alert?). One way to implement this is to compare strategy.totalAssets() and vault.strategies(address).current_debt for all yvUSD strategies on mainnet. If totalAssets < current_debt, the remote strategy has lost value and the next harvest will revert.

And for the existing edits, my AI claims to see some bugs:

1. CCTP alert label used for non-CCTP cross-chain strategies

Location: yearn/yvusd.py:328

Severity: Cosmetic

The staleness check on line 248 filters all strategies with type == "cross-chain", which includes
Katana (LxLy bridge) strategies, not just CCTP. But the alert message on line 328 always says
"*yvUSD CCTP Bridge Health Alert*". A stale Katana strategy would produce a factually incorrect
"CCTP" label in the alert.

The Katana yvUSDC Compounder (0xc5b16E7eFe1CA05714477b8edcAb4deE9b93a27C) uses
remote_chain_id: 747474 with LxLy bridging, confirmed in the live API response.

Fix: Change the alert message to something generic like "*yvUSD Cross-Chain Bridge Health Alert*"
or derive the bridge label from the strategy metadata (e.g., remote_chain_id).

2. remote_debt variable holds totalAssets, not debt

Location: yearn/yvusd.py:298, propagated through _build_cctp_alert_lines (line 430)

Severity: Semantic

_fetch_remote_strategy_state calls lastReport() and totalAssets() on the remote tokenized
strategy, returning (lastReport, totalAssets). The caller on line 298 unpacks this as
remote_last_report, remote_debt. The alert then displays this as "debt: $X".

For a remote morpho-looper, totalAssets represents the net equity of the leveraged position
(collateral value minus borrow), which can differ significantly from the mainnet vault's debt
allocation to that strategy. The staleness detection logic is correct — it only uses the timestamps
and the > 0 check — but the alert's "debt" label is misleading for responders comparing local
vs remote values.

Fix: Rename remote_debt to remote_assets or remote_total_assets throughout
check_strategy_staleness and _build_cctp_alert_lines, and update the alert text from "debt"
to "assets" or "position" for the remote side.

3. No dedup/debounce for CCTP staleness alerts

Location: yearn/yvusd.py:331

Severity: Impactful (alert spam)

Every other alerting path in the script has debounce logic:

• Flashloan liquidity: dedup key per chain.name + borrower address (line 617)
• Negative APR: dedup key per strategy address (line 217)
• APY inversion: alerted flag (line 195)
• Missing RPC: one-shot flag per chain (line 341)

The CCTP staleness check fires send_alert(HIGH, ...) on line 331 with no dedup. A bridge
outage lasting 3 days would produce 72 HIGH alerts (one per hourly run).

Fix: Add a per-strategy dedup cache key (e.g., YVUSD_STALE_ALERTED_{address}) that is set
when the alert fires and cleared when the staleness condition resolves, following the same pattern
used by the flashloan and negative APR checks.