# Which coding agent actually behaves like a pair programmer instead of an autonomous developer? — retained record

> Retained reddit record from reddit:ClaudeCode.

Canonical: https://highsignal.app/data/records/97745154402084e4

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:ClaudeCode)  published 2026-10-25 22:52 ingested 2026-10-26 02:39

[original source ↗](https://www.reddit.com/r/ClaudeCode/comments/1xfgbeo/which_coding_agent_actually_behaves_like_a_pair/)

## retained text



```
I'm looking for a coding agent that treats me as the architect and itself as the implementer. I want an actual collaborative relationship where it actively asks questions, presents decisions, challenges assumptions, and gets my input before making architectural or implementation choices.

I've been using GPT 6.1 Sol High with codex, and my biggest frustration is that it almost never asks questions. It regularly makes architectural decisions outside the scope of what I'm trying to accomplish, or introduces implementations that weren't necessary.

When I question those decisions afterward, the explanation is often something like "I made an assumption" or "I was accounting for" which it then acknowledges wasn't actually relevant and apologizes.

I've tried addressing this through AGENTS.md instructions and prompting, but it still tends to run off and make decisions independently.

It feels like these models are optimized for autonomy rather than collaboration.

Anyone have experience with agents or a harness that feels like this? Does Claude Code feel more autonomous or pair programming?
```



document fetched 2026-10-26 00:36 · canonical https://reddit.com/r/ClaudeCode/comments/1xfgbeo/which_coding_agent_actually_behaves_like_a_pair
