Show HN: Second opinion – A skill to query different models
github.comMake this skill to help me cross examine suggestions made by claude to other Claude or opencode supported models mid session. It compiles relevant context, consult a second opinion for me and support following up with the second opinion. A pattern that I find quite useful.
Currently it supports claude to claude, claude to opencode and opencode to claude invocation. These are the two harnesses I use.
This is a good idea - I've been manually doing something similar for almost a year now. It's interesting to find when GPT-5.5 will catch issues with Opus-4.8's code, and vice versa. Good thinking to make it a skill.
Thank you for the kind words! It is amusing that the first few second opinions tend to be all over the places, after back and worth with a few models, the area of contentions seem to settle. Having said that, I've also experienced none of the second opinions were addressing the actual issue.