Re:Don't be a meat proxy
by autumnstwilight on Monday August 03, 2026 @03:27AM
(#49152359)
I had an encounter with the opposite problem the other day, PM asked me if the library I was using for randomization was "secure" and I said, "Yes, I used the standard Java library for cryptographically secure random numbers that is already used in several other locations in this project," and he said, "But did you ask Claude if it's secure? It has to be secure!"
[ Parent
| on HN ]
Ask Claude to generate a few numbers using it and tell you if they are random or not
They may be random, but are they secure?
Re:Don't be a meat proxy
by antisthenes on Monday August 03, 2026 @10:41AM
(#49156447)
Just ask Claude if they are secure.
Re:Don't be a meat proxy
by mavelikara on Monday August 03, 2026 @04:33AM
(#49152869)
“I ask Claude only questions to which I don’t know answers to - which there are plenty of. That saves the company money by avoiding frivolous use of tokens.”
Re:Don't be a meat proxy
by EFreethought on Monday August 03, 2026 @05:02AM
(#49153095)
Asking someone to ask an LLM is peak laziness.
Beyond reverse centaur
Re:Don't be a meat proxy
by Viliam1234 on Monday August 03, 2026 @07:00PM
(#49162513)
Developers talk to LLMs.
Managers tell the developers to talk to LLMs.
CEOs tell the managers to tell the developers to talk to LLMs.
The longer the chain, the higher you are in the company hierarchy.
Re:Don't be a meat proxy
by thrownphone on Monday August 03, 2026 @09:23AM
(#49155507)
"Is Claude secure?"
Better get a Codex subscription to check!
Re:Don't be a meat proxy
by Obscurity4340 on Monday August 03, 2026 @02:49PM
(#49159869)
The cylinder must remain secure
Have your agent write a report on why it is secure, have it compare with other methodologies, pros and cons tables, and anything else you can pad out a few dozen pages, then ask the PM to review it and choose which method to follow before you proceed.
Repeat for all instances of micromanagement until they learn to trust your judgement and expertise.