Just launched Gruntend.
Give every task its own interface.
Gruntend lets LLMs compose your typed in-app functions into JavaScript plane, executed by a controlled interpreter while permissions and effects stay in your handlers.
Still beta. Try it:
https://t.co/nlWUAFoCy7
every memory solution I tried sucks, so I rolled my own
it is a plug and play prompt that gives any agent infinite memory
db is an append-only log. compression on the spot (nap, don't sleep). nothing is ever deleted, memory context is constant-sized, details fade with age
my personal testing experience isn’t so green either, even with agents, most of the times tests are n-1 times correct and then you find that one failing case and now you have n+1 tests
It could simply be survivorship bias that I don’t see that tests that don’t fail blah blah
pretty sure I’m not the first one to say this but why ci/cd taking so much time is bottleneck. it takes 15 minutes avg to decide whether it can be deployed, are people doing anything different?
It would never get here if a disappointed customer hadn’t called me last year nov and said that the software is shit. that was damaging to my reputation and we had to recapture the ground. We did.