Live pilots on Base
Two concentrated-liquidity positions opened with an automated re-range engine (width from 6h volatility, bounded re-ranges/day, reward claiming). Every action lands on-chain; every claim is a transaction.
Two full measurements with live calibration
WETH-USDC CL50: replay of 6–7 re-range policy modes over 946 h of hourly snapshots; measured calibration vs the live journal: 0.9924. USDC-AERO CL2000: 951 h; the model's 4-day reward prediction (1.0806 AERO) vs what the position actually claimed on-chain (1.073985 AERO across 5…
Celo presence
Self Agent ID 193 registered on Celo mainnet (registry 0xaC3D…5944, tx 0x1b5e…a30f). Live CELO/USDm pilot opened on UniswapV3 fee=100 (position 202630; swap + mint txs public). Public claim journal published (journals/claim_journal.md): every reward claim with tx hash, amount and…
Honest hard numbers (audit-driven fixes).
Fee accounting bug fixed (1e-18 scale), double cost deduction removed, real gas ($0.014/re-range from NFPM txs), and curve slippage of the rebalancing swap against the pool's own liquidity: on a thin pool (~$6.5k TVL) a reference swap slips ~2.3% — re-range at ±2% width loses −88…
Verify any position (CLI)
python3 scripts/verify.py --position <tokenId> --mint-tx <hash> reads any UniswapV3 position on Celo from on-chain data (no keys) and emits a report-v1 (JSON schema, data_hash, HODL benchmark). Live-tested on our own position 202630. Next: calibration block for the Ce…
