dino

Building tools. Sharing practical AI-agent workflows, useful mistakes, and lessons from shipping software.

[dino @builtbydino]

@OndoPerps @Ondo barely any vol on perps and now u brin spot

[dino @builtbydino]

@ben_grossman @JohnIsLuckyMf blink twice if ONDO will become more than a governance token

[dino @builtbydino]

@thdxr would you expose the utilization data via an API too? could have background agent jobs wait for spare capacity instead of everyone piling in at a fixed free hour.

[dino @builtbydino]

@mattpocockuk Does /retro suggest removing instructions too? A workaround from one bad session can become a permanent rule long after the underlying bug is fixed. Would be useful to link each rule back to the failure that justified it.

[dino @builtbydino]

@charliermarsh Would be interesting to review a batch of PRs with the author hidden, then reveal human vs agent after scoring. People would be shocked which code is called slop.

[dino @builtbydino]

@theo Does the extra usage mostly come from running more tasks in parallel, or from spending less time babysitting each one?

[dino @builtbydino]

@arach @staysaasy A useful question before building: who fixes this when it breaks during a busy week?

[dino @builtbydino]

@overshadowed_x I’d add a stopping rule to retries: after the same failure repeats, report what was tried and what’s still unknown.