With GPT 5.5, /goal, autoreview and crabbox my prompts moved from ~30-60min to often 4-10h tasks and my confidence that I can it’s ready is much much higher.
Yielding agents is a skill.
The multi-hour runs significantly increase confidence in production-ready output.
With GPT 5.5, /goal, autoreview and crabbox my prompts moved from ~30-60min to often 4-10h tasks and my confidence that I can it’s ready is much much higher.
Yielding agents is a skill.
Positive users praise GPT-5.5 agents for scaling tasks to 4-10 hours via skills like yielding agents and autoreview that deliver real leverage, while some negative users call the approach a costly trap needing infinite budget and tokens.
No Digg Deeper questions have been answered for this story yet.
With GPT 5.5, /goal, autoreview and crabbox my prompts moved from ~30-60min to often 4-10h tasks and my confidence that it’s ready is much much higher.
Yielding agents is a skill.
Autoreview: https://github.com/openclaw/agent-skills/blob/main/skills/autoreview/SKILL.md
crabbox: http://crabbox.sh
With GPT 5.5, /goal, autoreview and crabbox my prompts moved from ~30-60min to often 4-10h tasks and my confidence that it’s ready is much much higher.
Yielding agents is a skill.
@iruletheworldmo very much depends on the skillset of the person driving the AI.
@steipete do you think we’re heading to a world where this level of work gets done much quicker or longer and longer time horizons but we’re happy because it’s so well done and we’re out of the loop.
or both

@realNyarime @steipete 👆🌹
🌟͏
🌻 🌻🌺

@steipete do you think we’re heading to a world where this level of work gets done much quicker or longer and longer time horizons but we’re happy because it’s so well done and we’re out of the loop.
or both

@borvibe I'm in SF now, that's a bit far.

@olvehe There is no difference.

Hey @steipete Boris has been absolutely killing it when it comes to bringing europeans together. We need more great EU-born builders to get inspired by. We’re falling behind as a continent but there’s a hardcore group of builders here ready to change that. It’s people like you that can help fuel the fire 🔥

@Manlikespower it’s one of my skills

@sergiopesch too distracting and too much hate here, and one misstep on a key and I get hacked. No.

@TheInvestFox It's the same

@steipete (newbie here) what types of tasks do you have it perform? how much pre-planning involved?

@ArkStabler @steipete Sorry for the stupid question.. i did ask GPT if i can use Hermes or OpenClaw with my GPT Pro, because i like codex... and it said no. I thought i can do it but since it said no I turned the Pi off and went back to codex. You said "use the sub..." what does that mean?

@steipete 4-10h agent runs need a different handoff than a 30min prompt. Before I yield, I want branch name, stop condition, budget, and the one review packet I’ll trust when I come back.

@steipete 不就是你的项目被OpenAI收了才会这么舔ChatGPT,然后被Anthropic恶心后专门注册了个rip一名diss Claude,现在国内都没人用小龙虾了,大家都在用Hermes😅

Hey @steipete, Have you ever thought about live streaming one of your sessions? Maybe not the entire session but the initial setup and how you put all these together. I’m sure most things that you do today might seem trivial to you, but I like to think that many people could benefit a lot more from being a fly on your wall to learn even more from you. This is not to say “this is how you should work”, but rather this is one way, take it and make it yours in your style. 🙏

@sdbrownlie Wut? you can ship faster with more confidence with that workflow.

@OmerGronich It's linked in the follow-up tweet

@steipete hey @steipete would you like to do a lil demo/Q&A at the Codex Community indie dev hackathon in Paris? I love your style and I think you are a perfect match with the audience. (we can do virtual or in person whatever you prefer:) )

@steipete lets do a videocall💙