List view
We have a general imperative to fuzz the protocol and find as many bugs as possible - [ ] invariant checks during fuzzing - [ ] automate deployment of infra fuzz pipeline - [x] automate crash report delivery & notification - [x] make it as easy as possible to troubleshoot bugs found during fuzzing - [x] reduce false positives - [x] write more unit tests around fuzz pipeline to make it easier to narrow down bugs - [ ] load bot & chain state in python to ensure the bug is in solidity - [x] add expected anvil errors that we can ignore/retry around w/ warnings
Overdue by 1 year(s)•Due by February 21, 2024•5/5 issues closedThe goal is to demonstrate smarter fuzz bots for identifying economic bugs in Hyperdrive. Deliverables: - A proposed code structure for an RL-compatible environment "gym" that facilitates trading on Hyperdrive. - A demo (simple) RL model that can learn to make sensible (more profitable than random) trades on Hyperdrive. - [x] train a model - [ ] script to launching it as a bot on devnet - [ ] merge code into monorepo - [ ] write smarter fuzz bots for identifying economic bugs
Overdue by 1 year(s)•Due by March 22, 2024•0/3 issues closed