Keepers

Keepers are off-chain services that call permissionless contract functions to keep the protocol running. They are not trusted — anyone can run a keeper, and the protocol works correctly regardless of which keeper submits a transaction.

Batch Clearing Keeper

Calls clearBatch() on active markets at the configured batch interval.

• Monitors pending order volume via segment tree reads

• Skips clearing if no crossing orders (saves gas)

• Gas estimation + exponential backoff on failures

• Multi-market scheduling: round-robin or priority by volume

# Configuration
BATCH_KEEPER_RPC_URL=https://...
BATCH_KEEPER_PRIVATE_KEY=0x...
BATCH_KEEPER_INTERVAL_MS=3000
BATCH_KEEPER_GAS_LIMIT=2000000

Resolution Keeper

Resolves expired markets with Pyth oracle data.

• Watches for markets entering Closed state

• Fetches signed update data from Pyth Hermes: GET /v2/updates/price/{publishTime}?ids[]={priceId}

• Submits resolveMarket() then finalizeResolution() after finality window

• Handles fallback window extension if initial fetch has no data

• Claims resolver bounty on success

Pruning Keeper

Cleans up expired orders to reclaim storage.

• Scans for expired/stale orders (via indexed events or contract reads)

• Calls pruneExpiredOrders(orderId[]) in batches

• Claims pruner bounty (order bonds)

• Runs less frequently (~30s intervals)

Deployment

All keepers run as systemd services:

# Install
cd keeper && npm install && npm run build

# Systemd
sudo cp deploy/systemd/strike-*.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl enable --now strike-batch-keeper strike-resolution-keeper strike-pruning-keeper

Monitoring

• Keepers log all actions with timestamps and gas costs

• Alert on: missed clearing intervals, failed resolutions, RPC errors

• Health endpoint: GET /health returns keeper status and last action timestamp