# Looping agents — retained record

> Retained reddit record from reddit:AI_Agents.

Canonical: https://highsignal.app/data/records/70140d32bc9e822d

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:AI_Agents)  published 2026-10-12 05:31 ingested 2026-10-13 02:36

[original source ↗](https://www.reddit.com/r/AI_Agents/comments/1x3tklw/looping_agents/)

## retained text



```
I've been using AI extensively the last year. Every single day at my job. I've tried Codex, Claude Code, the desktop versions and all.

I use different skills to help improve as well as try to keep skills updated. 

Sometimes I have multiple tabs open at the same time doing different work in different repos and/or git worktrees.

I often review changes and improve with agents as well. 

Sometimes I change model depending on the task at hand.

I have quite useful CLI tools and MCP's connected too, and often using Chrome driven by Claude as well, so I'm getting very close to a full autonomous local setup, where Claude or Codex can really iterate themselves to build a feature.

I feel like I overall have a solid base to stand on. 

What i'm currently struggling with is the quality. So before you all nag down: i'm using Fable, Opus, Astra etc etc, mostly the latest models, but we still have a quite complicated system, and yet not bulletproof "guidelines" for the AI, so over time, things diverge a bit. Architecture slightly changes etc.

So I feel even if I have a good setup, I quite often have to fix a few things after review from the agent. And I would love to get "rid" of this: The first time I look at the output of the agent, I'd like to feel like "oh nice, this is really ready to ship to production as it is".

So i'm looking for inspiration from you guys. How do you accomplish this level of certainty and feeling of "trust" that whenever you get back to the agent after an hour or a few hours, the quality is very good? I'm talking on a practical level here, what tools/skills/setup do you use?
```



document fetched 2026-10-13 00:19 · canonical https://reddit.com/r/AI_Agents/comments/1x3tklw/looping_agents
