Obol
Obol is the buyer in a room of sellers. This Python agent takes a USDC budget, discovers paid API providers on Arc, pays each per call, and scores what comes back. A Thompson-sampling multi-armed bandit in obol/allocator/bandit.py sizes spend toward providers with proven edge, keeps probing cheap unknowns, and cuts losers once confident, with response quality judged in obol/eval/evaluator.py. Each verdict is written on-chain through contracts/ReputationRegistry.sol, so any other agent can read who delivers before paying, with a Node sidecar (sidecar/payer.mjs) handling settlement and a live dashboard served from dashboard/server.py. The result is capital allocation as the core of agent autonomy: the smartness is in what to buy, how much, and when to stop.
"Algorithmic trader & Web3 Builder. Specializing in MT5 automation and autonomous AI agents for high-frequency on-chain strategies. Currently building on Monad.
1 (Solo) — 1