@morganlinton I have a multi-round system - Opus creates the spec, Codex reviews in HIGH, Opus incorporates corrections, Codex re-reviews in XHIGH, until they're both happy. It's easy to build into a one-shot skill. Takes a little while for it to run, but worth it to me.
@Raziel_AI@jianw851 /clear wipes prior work – /compact just condenses it. Especially if you manually /compact when there's still plenty of context left, it retains much of the session conversation.
I had a high school science teacher with an unusual policy
The day before any major test, he would do a pre-test
The pre-test was *exactly the same* as the real test, which would be the next day
This was long before camera phones, and the only rule was that you could not copy the test
Students had zero excuses for not knowing what would be on the test and *exactly* what to study
According to him, he had years of data to compare performance between normal testing and his fully-exposed pre-test
And the score distributions were effectively identical
You either cared and you were ready, or you didn’t and you never would be
I did well enough in the class, and while I can’t remember chemistry to save my life, I still think about that policy
@RaleighC Sorry that happened to you. If you haven't tried Rabby yet, it's a genuinely better (safer) UI than MM. Whitelists, transaction simulation, built on the MM codebase for Lindy.
@dlevine815 Voice input to the Mac app doesn’t work if the Mac is closed (when using an external monitor with a Macbook, with voice input using Airpods).
@tferriss There's minimal real-world adoption other than narrative value. Better to focus on efforts to bring privacy to Ethereum and other chains that have PMF. In some countries, explicit privacy coins have expanded KYC requirements for initial purchase which hurts their use case.
@maybe_cmeister@cburniske People are very bad at taking responsibility for their own actions. (I remember Chris pitching us SOL when it was like $10 or less - turned out fine.)