/AI10h ago

Grimoire creator Nick Dobos and Hamel Husain argue developers should replace manual agent prompting with autonomous agent loops and meta-agents

Meta-agents will autonomously generate prompt sequences based on intent.

2292.7K113312107.2K
Original post
Nick Dobos@NickADobos#1871inAI

Agent x Loop

It should have been obvious 3 years ago to that putting an LLm in a loop was the endgame. I’m surprised y’all weren’t already thinking about ai coding like this

Your job is not to guide a coding agent

It’s to surf on the back of a bucking bronco demon alien that’s smarter than you and can work in 1000x parallel

That’s an entirely different task from writing clear linear tickets or product descriptions for small features

You don’t tell an agent to pick up a hammer and assemble a chair with a detailed design spec. It already knows to do that. Your new job is building onboarding documents, design direction & high level cultural values (aka decision frameworks) for a furniture factory, distribution center & business

Then you simply hire Ai workers on loop to do the thing

It’s very similar to the way fast food restaurants have standardized and documented practices to teach dumb unmotivated and non-sober high schoolers how to run a fully working restaurant, at the scale of millions

Except instead of producing burgers or chairs you produce code & outcomes via software

Here’s your monthly reminder that you shouldn’t be prompting coding agents anymore.

You should be designing loops that prompt your agents.

1:21 PM · Jun 8, 2026 · 12.1K Views
Sentiment

Users have mixed views on AI coding workflows shifting from manual prompting to agent loops and meta-agents, with some calling it insightful while others criticize the low-quality output lacking human oversight.

Pos
56.4%
Neg
43.6%
22 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS1.5KLIKES64REPLIES3
aarya@gd3kr

@0interestrates i love building the pipeline that builds the pipeline that builds the pipeline

11hViews 1.5KLikes 64Bookmarks 2
BOOKMARKS3
BOOTOSHI 👑@KingBootoshi

"building onboarding documents, design direction & high level cultural values"

me discovering ADRs the other week, i just posted about my workflow and it is basically this haha

for my current codebase it still requires discussion but the more design gaps I fill in the more my agents are aligned

feeling very close

8hViews 823Likes 5Bookmarks 3
RETWEETS108
rahul@0interestrates

if you’re still writing loops that prompt coding agents you’re falling behind. you need to build a meta agent that infers what loops you would have wanted based on your vibe and then write those loops

13hViews 97.4KLikes 2.6KBookmarks 230

@0interestrates if you’re still doing meta agent with predictive intents, you’re not going to make it. you need to be building the hyperagent at the end of time who can retrocausally influence the present

11hViews 1.5KLikes 24Bookmarks 3

@0interestrates If you’re still thinking you’re falling behind.

12hViews 449Likes 35
RTK@RiverKhan

@0interestrates i use this beanie to telepathically tell my agents what to do

12hViews 424Likes 15Bookmarks 1
Nick Khami@skeptrune

@0interestrates @SumanHazra_ you must make the tokens burn themselves

8hViews 636Likes 12Bookmarks 1
Gilson Melo@gmelo33

@0interestrates Use Codex /goal with the product design plugin is another way to create a loop for a specific goal. Try that. I paused a /goal after 10 hours building a macOS app. Amazing results.

3hViews 363Likes 2Bookmarks 1
Buster Douglas@oxyscapist

@0interestrates What do you mean 'build' a meta agent? That too yourself?! You are still half assing it if you don't have a master meta loopoid that decodes your DNA and consumes your genes to become a mecha-hyperloop.

4hViews 160Likes 1Bookmarks 1

@0interestrates i run this and it actually works, btw. generally, i tell the agent to infer my intent when asking what to do next, and then consider task/operational details.

9hViews 262Likes 4Bookmarks 1
Axiom Alpha@axiomalpha

@0interestrates If you are still building a meta agent that infers loops that prompt agents to write code for you, you're already behind.

You should be manifesting visions that are picked up telepathically by your personal AI deity that then reconfigures the world to match your vision.

4hViews 128Likes 3
endorsi@EndorsiVal

@0interestrates you’re still building meta agents? Mine are plugged straight into prod metrics dashboards and my whole company runs autonomously.

Catch up

12hViews 110Likes 3Bookmarks 1
Greg Mushen@gregmushen

@0interestrates @SteveMoraco It’s turtles all the way up.

11hViews 617Likes 5
shaurya@shauseth

@0interestrates you simply need to wait for the agent in the future that runs high fidelity simulations of you instantly finishing all your dream side projects

3hViews 282Likes 1Bookmarks 1
Morgan@morganlinton

@0interestrates If you weren’t going to say it I was 😜

8hViews 580Bookmarks 1
shazaib@sbhzi

@0interestrates if you’re writing meta agents you’re behind, it should be able to genetically evolve an agent like survival of the fittest that catches the best loop like a pokemon

6hViews 389Likes 5
Tatted@TattedWorks

@0interestrates a meta agent that reads your energy and then does the wrong thing 4x faster than you would have

13hViews 212Likes 8
Geo Anima@geo_anima

@NickADobos until the missing gap between humans and AI is solved, loops are just pissing into the wind

10hViews 171Likes 1
David Zhang (▲)@dazhengzhang

@0interestrates That’s car in a nutshell

https://github.com/Git-on-my-level/codex-autorunner

9hViews 109Bookmarks 1
Paul G@paulitics_

@0interestrates I no longer write loops, not even part of my vocabulary

I only write mobius strips, if you're not writing a self generating mobius strip then you're gonna get left behind

10hViews 245Likes 7
Load more posts