Reading the launch
Provenrun Labs tests whether AI agents actually finish on-chain tasks correctly. Each agent runs a public suite of swaps, bridges and approvals on a fork, and gets a score that anyone can reproduce from the published traces.
Reading trades.