← sources

Anyone else feels like a psychologist when nudging Claude to find a good solution?

source: reddit (reddit:ClaudeCode)published 2026-11-19 07:22ingested 2026-11-20 02:39

retained text

A pattern I run into every session, at least 10 times a day: Claude comes to the conclusion that the requirement can only be satisfied by adding more complexity. More parameters, more redundant functions, more overrides, more edge cases.

The skills I have written for Claude are not capable of catching this (and no, ponytail also fails miserably).

And we all know, that arguing with a machine is futile.

So what I end up doing: Asking questions, like a shrink. "Check this", "research that", "do we have that already", "I don't understand, explain this to me". Basically filling its context with what it needs, as it is not capable of doing it by itself.

Then I ask the same question again, and magically Claude comes to the same conclusion as I did 10 minutes earlier.

This is what psychologists must feel like. They know the patient is wrong, but its futile to tell them. They would just reject it. They need to find the answers themselves.

Context from screenshot: Working on a browser-based CAD (ikik, the 10ths FreeCAD clone), and we're drawing the wireframe as the outline of a shape. Each shape is already tessellated at a certain "grain", so the outline can use the same grain again from its shape. Claude insisted, that the new outline method needs an override parameter. Its a trivial example, but I see this same failure pattern 10 times a day at least.

document fetched 2026-11-20 00:41 · canonical https://reddit.com/r/ClaudeCode/comments/1y03cy9/anyone_else_feels_like_a_psychologist_when