Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Don't Build Multi-Agents

cognition.ai

Post reply on HN
Print view View original

91–95 of 95 posts

Re: Don't Build Multi-Agents

#91

Post by worik »

adastra22 wrote: post #80
worik wrote: post #60

Earlier quoted context omitted.

One reason it is concerning. I am mostly worried that I am wrong, in my opinion, that "agents" is a bad paradigm for working with LLMs I have been using LLMs since I got my first Open AI API key, I think "human in the loop" is what makes them special I have massively increased my fun, and significantly increased my productivity using just the raw chat interface. It seems to me that building agents to do work that I a…

The thing is, the same agent that made the bananas mistake is also quite good at catching that mistake (if called again with fresh context). This results in convergence on working, non-bananas solutions.

Look up The Old Lady who Swallowed a Fly. Or The King, the Mice and his Cheese

What you propose makes things worse, not better

LLMs are magnificent tools, but there needs to be a human hand holding them.

Nothing I have seen anywhere, yet, challenges my view that "agents" will not be a good idea until we have better technology, that there is no sign of yet (?), than LLMs.

Re: Don't Build Multi-Agents

#92

Post by kikuska »

clbrmbr wrote: post #44
adastra22 wrote: post #11

> As of June 2025, Claude Code is an example of an agent that spawns subtasks. However, it never does work in parallel with the subtask agent, and the subtask agent is usually only tasked with answering a question, not writing any code. Has this changed since June? Because I’ve been experimenting over the last month with Claude Cide subagents that work in parallel and agents which write code (doing both simultaneousl…

I’ve been quite successful since June doing parallel edits just on different components within the same codebase. But I’ve not been able to do it with “auto-accept” because I need a way to course correct if one of the agents goes off the rails.

Forgive me I can't figure out how to reply on the original thread but I saw you used a script to build an Anki deck with the native plant species in your area. I'd love to do the same but I've never coded in my life and doing it manually would take forever. Would you be willing to share how this is done? Thanks!

Re: Don't Build Multi-Agents

#93

Post by clbrmbr »

kikuska wrote: post #92
clbrmbr wrote: post #44

Earlier quoted context omitted.

I’ve been quite successful since June doing parallel edits just on different components within the same codebase. But I’ve not been able to do it with “auto-accept” because I need a way to course correct if one of the agents goes off the rails.

Forgive me I can't figure out how to reply on the original thread but I saw you used a script to build an Anki deck with the native plant species in your area. I'd love to do the same but I've never coded in my life and doing it manually would take forever. Would you be willing to share how this is done? Thanks!

Ooo so glad you asked. That's something I am super passionate about. I built those decks during the COVID lockdowns when I needed to connect with nature a bit more. Those Anki decks were incredibly efficient for learning to recognize many plants and mushroom species!

I actually did very little coding. It was more of like just some light scripting and today with LLMs you should be able to figure it out even without coding experience.

Here's how I did it:

1. I fetched information on the 500 most commonly observed plant species in my area (Northeastern USA and Canada) on iNaturalist. I think I did this just by downloading the web pages with wget. inside of those downloaded files were the common and scientific name of the species as well as a link to the photo.

2. I then wrote a very simple script to download all of those images.

3. Finally, I had a script that I think just created a CSV file that I was able to load into Anki and I created my template, making sure to have a notes section so that as I'm studying, I add interesting facts, such as the genus Eupatorium being named after an emperor infamous for his use of poisons [1].

A note on copyright: Although I would have liked to have a deck that I could freely share with people on the internet, I really wanted to have the best possible photos to help with identification. So I chose to use the primary photograph associated with each species on iNaturalist and most of these are copyrighted without any Creative Commons license. So it's totally fair use to use it for your own edification, but I'm unable to share one of these decks with you directly without getting permission from hundreds of iNat contributors.

[1] https://en.wikipedia.org/wiki/Mithridates_VI_Eupator

Re: Don't Build Multi-Agents

#94

Post by kikuska »

clbrmbr wrote: post #93
kikuska wrote: post #92

Earlier quoted context omitted.

Forgive me I can't figure out how to reply on the original thread but I saw you used a script to build an Anki deck with the native plant species in your area. I'd love to do the same but I've never coded in my life and doing it manually would take forever. Would you be willing to share how this is done? Thanks!

Ooo so glad you asked. That's something I am super passionate about. I built those decks during the COVID lockdowns when I needed to connect with nature a bit more. Those Anki decks were incredibly efficient for learning to recognize many plants and mushroom species! I actually did very little coding. It was more of like just some light scripting and today with LLMs you should be able to figure it out even without co…

Yay this is awesome thank you!! I will try this out :) cheers for the detailed guide, had no idea about wget - that will help a ton!

Re: Don't Build Multi-Agents

#95

Post by kordlessagain »

adastra22 wrote: post #75
kordlessagain wrote: post #62

Earlier quoted context omitted.

I wrote something that watches the directories and started working on a node graph to visualize relationships between changes and dates things like that: https://github.com/kordless/gnosis-flow . It that is useful to you let me know!

Isn’t it easier to just write scripts to launch clause in different worktrees?

I have no idea what you are saying.
Post reply on HN