Insurance Fund

In one sentence: a reserve that quietly skims a small slice of treasury inflows to build a safety buffer, and automatically stops topping up once it is big enough relative to the protocol's size.

The insurance fund exists to absorb future losses — for example, reimbursing the pool after a validator slashing event. It fills itself passively from protocol fees and is sized as a fraction of how large the protocol has ever grown.

How it fills

On every reward sweep, after the protocol fee is split, a fraction of the treasury slice is diverted into the insurance pool (canoliq/insurance/pool):

insurance = treasury_slice × insurance_bps ÷ 10,000

• insurance_bps default: 500 (5% of the treasury slice). Since the treasury is 30% of the fee, this is roughly 1.5% of the total fee — consistent with the whitepaper's "small fraction of treasury inflow" framing.
• The skimmed amount is taken out of the treasury credit, not added on top, so fee conservation still holds exactly (see below).

The target-based auto-off (T4)

A reserve that grows forever is wasteful. canoLiq caps the fund relative to peak TVL — the largest the pool has ever been — and switches the skim off once the target is reached:

target = peak_tvl_ucnpy × insurance_target_bps ÷ 10,000

if insurance_target_bps > 0 and insurance_pool >= target:
    insurance = 0    # skim turns off; the slice stays in the treasury

• insurance_target_bps default: 500 (5% of peak TVL).
• Setting insurance_target_bps = 0 disables the gate entirely, so the skim is always on.
• Peak TVL only ratchets upward. It is updated each block as peak_tvl = max(peak_tvl, total_pooled_cnpy). When the protocol grows to a new high, the target rises with it; if the fund now sits below the new target, the skim automatically resumes until it catches up.

This makes the reserve self-balancing: it fills to ~5% of the protocol's high-water mark, pauses, and tops up again only as the protocol grows.

Fee conservation

Whether or not the skim is active, every component of a reward sweep must sum exactly to the reward delta — the would-be insurance amount simply stays in the treasury when the skim is off:

user_rebate + treasury + insurance + validators + buyback = fee amount

Any test asserting conservation must include the insurance line item.

Monitoring

The /v1/pools endpoint exposes the current balance, the computed target, and how funded the reserve is:

GET /v1/pools
{
  "insurancePool": 4200000,
  "peakTvlUcnpy": 1050000000,
  "insuranceTargetUcnpy": 52500000,
  "insuranceFundedBps": 800
}

insuranceFundedBps is insurance_pool ÷ target in basis points (here, 8% funded).

Status

Phase	Usage
Now	Passive accumulation with target-based auto-off (T4)
Future	Slashing-reimbursement disbursement

Related

• Fee Structure — where the treasury slice comes from
• Push Alerts — the tvl_drop alert that warns of the events insurance covers
• API Endpoints — /v1/pools