# How PMs use the Codex app

## Метаданные

- **Канал:** OpenAI
- **YouTube:** https://www.youtube.com/watch?v=6OiE0jIY93c
- **Дата:** 09.02.2026
- **Длительность:** 2:20
- **Просмотры:** 26,291
- **Источник:** https://ekstraktznaniy.ru/video/11115

## Описание

Alexander Embiricos (a Product Manager on the Codex team) shows how he uses Codex skills to make a small product change, diagnose a Buildkite failure, and improve the skills so the next PR goes faster.

Takeaways:
- Skills are a shortcut for repeated workflows like Buildkite logs.
- When a skill fails, fix the root cause and update the skill.
- The real win is compounding: the codebase gets easier over time.

This is the loop: ship the fix, then teach the workflow.

Chapters:
00:00 PM context: careful before tagging engineers
00:16 A confusing button and a quick team check
00:32 Delete the button, then hit a PR failure
00:46 Use the Buildkite skill instead of digging through logs
01:09 Install the Buildkite token
01:20 Update the skill so it works next time
01:53 Inductive loop: feedback - fix - improve the skill
02:06 Compounding payoff: Codex gets better over time

## Транскрипт

### PM context: careful before tagging engineers []

As PM of the team, I don't code that often, so I like rarely understand the code that's being edited, like and the code around the code that's being edited. So I kind of want to read up on that before I look stupid in front of the team. I actually want to be kind of careful, like when you decide to tag any engineer or even when you want to file a ticket. In this case, I saw, you know, there was this

### A confusing button and a quick team check [0:16]

button and it was kind of confusing as to what it did. So my gut, my immediate thing is to say, okay, what does this button even do? I'm just asking a question. So then I pinged the team and I was like, hey, this is a strange behavior to me. I just confirmed that we don't need this button. You know, I don't want to go around deleting people's buttons for no reason. So like, oh, we don't

### Delete the button, then hit a PR failure [0:32]

need it. Okay, great. So I told it to delete it. And actually I got distracted and I was working on some other stuff. Saw the change. We went back and forth. So I created PR and I saw that my PR got approved, but there was a test failure. And, you

### Use the Buildkite skill instead of digging through logs [0:46]

know, I have become so lazy. Like the moment I'm like, oh, I better go into like build kite and like, look at the test logs. You know, it's like, ah, yeah, but the app has all these great skills. And so I could just click into the skills tab and I'd just be like, build kite. Okay. Sick. Someone is like already figured this out for me. So I'd be like, okay, let's go back to this thing and just be like build kite fetch logs. Why failing? And it told

### Install the Buildkite token [1:09]

me, okay, you need to install the build kite tokens. So I did that. So I was like, okay, I ran the command to get a token now fix my PR. So what I immediately did is I went to our main

### Update the skill so it works next time [1:20]

the top level folder in the repo, which is where all the skills are. And I just asked it to fix this problem. I had to ask you how to set the build kite token. I don't want you to have to do that next time. So, you know, read the rollout, reflect on what might fix it and like update the skill. I probably could have just said update the skill, but I feel like kind of helping it, like know how to think through it results in higher quality, especially because I'm often using codex on low, which I think is like an under underrated technique. And so basically we, you know, we went, had a bit of a back and forth, but ultimately made an update to the thing that is in a PR somewhere. Just to summarize, like we

### Inductive loop: feedback - fix - improve the skill [1:53]

went through this inductive journey where I was looking at user feedback and I saw a thing, I asked the question, realized this probably needed to be changed. So then I like made the change in, and then in making the change at a PR failure. So, so

### Compounding payoff: Codex gets better over time [2:06]

I used a skill to fix the PR failure, but in the mean, then improved two skills kind of as I went. And so over time, codex just gets like better and better in working on a code base, which is pretty fun.