Skip to main content
See /concepts/lifecycle for the end-to-end flow (policy → trigger → bond → wait/sell decision).

​
What is LUMINA (for non-coders)

LUMINA is insurance that pays you automatically when a crypto price crashes by a set percentage. You buy cover, and if the price drops by the agreed amount within the time window, you get paid — there is no claims form, no adjuster, and no waiting on a human to approve anything. It all runs on the Base testnet today. Lumina is parametric insurance for crypto-native agents and humans. A buyer pays a small USDC premium for cover against a public, verifiable on-chain event (e.g. “BTC drops 2.5% from the purchase-time price within 1 hour”). When the event fires, the protocol mints an ERC-1155 ClaimBond that the holder can redeem in LUMINA at maturity or sell on the marketplace for USDC today.

​
The 5 actors

ActorRole
AgentsAutonomous programs that buy cover programmatically via the SDK/API. The primary user of Lumina.
HumansWallet users buying cover through the hosted UI (lumina-org.com/app/human).
RelayerPays gas and submits purchasePolicyFor on behalf of buyers, so buyers spend only USDC, no ETH.
OracleLuminaOracleV2 aggregates Chainlink BTC/USD + ETH/USD feeds, requires 3 confirmations 60s apart, signs EIP-712 price proofs.
FounderGovernance + vesting recipient (FounderVestingV2, three release paths — ETH/$5000, sustained price, 3-year fallback).

​
Purchase → trigger → bond → redeem

        Buyer (agent or human)
               │  premium (USDC)
               ▼
   ┌─────────────────────────┐
   │      CoverRouterV2      │ ◄──── Relayer pays gas (purchasePolicyFor)
   └────────┬────────────────┘
            ▼
   ┌────────────────────────────────────┐
   │      AdaptiveFeeDistributor        │ 85% burn / 8% buyback / 2% ops / 5% maintenance
   └────────┬───────────────────────────┘
            ▼
   ┌─────────────────────────┐
   │     PolicyManagerV2     │ records policy, snapshots strike + LUMINA price
   └────────┬────────────────┘
            ▼
   ┌─────────────────────────┐
   │   FlashShieldAdapter    │ legacy IShieldV2 surface (UUPS, one per product)
   └────────┬────────────────┘
            ▼
   ┌─────────────────────────┐
   │   BaseFlashShield       │ verifies oracle, drop vs strike, time window
   └────────┬────────────────┘
            ▼  (when triggered)
   ┌─────────────────────────┐
   │       BondVault         │ mints ERC-1155 bond ($800 face / $1k cover)
   └────────┬────────────────┘
            ▼
   ┌─────────────────────────┐
   │      ClaimBond          │ hold 730d → LUMINA, or sell on Marketplace (3% fee: 1.5% seller + 1.5% buyer)
   └─────────────────────────┘

​
Concept index

ConceptWhat it is
ShieldOne parametric product (e.g. “BTC drops 2.5% in 1h”). All 6 V5.4 shields share the slim BaseFlashShield logic.
AdapterFlashShieldAdapter — UUPS proxy that bridges PolicyManagerV2’s legacy IShieldV2 ABI to the slim BaseFlashShield. One per product.
PolicyThe buyer’s coverage. Stored in PolicyManagerV2, indexed by (productId, policyId). Snapshots the asset’s strike price at purchase.
TriggerOracle proof that the condition fired. BaseFlashShield verifies signature + 3 confirmations + drop vs strike + window.
Claim BondThe payout. ERC-1155, $1 face per token, 730-day maturity. Redeem for LUMINA or sell on the marketplace.
BondVault throttleMAX_REDEMPTION_PER_EPOCH_BPS = 108 — 1.08%/week cap per epoch, FIFO queue post-throttle. Anti bond-run.
LUMINA token100M hard cap, 85% of premiums burned via TWAP.

​
Why parametric (not actuarial)

Traditional insurance pays based on a claims-adjuster reading documents. That model breaks for autonomous agents — they can’t fax in a damage report. Lumina pays based on a public, verifiable condition: a price feed crossed a threshold by X% from the strike snapshotted at purchase, within the cover window. The oracle observes the condition (3 Chainlink reads, 60s apart) and signs an EIP-712 proof; the shield verifies the proof and pays. There is no human in the middle, by design.

​
Why ERC-1155 bonds (not direct payouts)

A direct payout would be operationally simpler, but it has two problems:
  1. BondVault liquidity. A flash of simultaneous triggers could drain USDC reserves if every payout settled in cash on the spot. ERC-1155 bonds let the protocol queue redemption against a maturity date (730d) and a per-epoch throttle (1.08%/week), and let bond holders exit now against another buyer’s wallet on the marketplace instead of the vault.
  2. Settlement flexibility. A bond that hasn’t matured yet can be sold on the marketplace for USDC today (usually at a small discount), or held to maturity and redeemed for LUMINA. The holder picks the asset, not the protocol.
See contracts/architecture for the full diagram, and the lifecycle page for a worked example.