#agent

Hacker Newsh4ckernews
2025-06-19
2025-06-17

When working with #coding #agent do they commit under their #llm?

#git

2025-06-16

Claude 3.5 Sonnet в роли AI-агента: тестирую работу в цикле с реальным выполнением кода

Вчера тестировал мой агент DepthNet на реальных моделях. Это что то вроде "операционной системы для LLM моделей" В прошлой статье пользователи просили привести диалоги, и рассказать подробнее, и тогда я решил более полно осветить вопрос.

habr.com/ru/articles/918864/

#agi #agent #llm #llmагент #php #laravel

David J. Atkinsonmeltedcheese@c.im
2025-06-15

@w7voa 4/ Lying Machines (cont.)
In his # #speculativefiction book, “The Windup Girl,” writer Paolo Bacigalupi tells of a world where genetically-modified human robots are indistinguishable (to our senses) from ordinary humans and this is a problem. A rule is made that all such entities are compelled to signal their non-human status through easily observed jerky movements. There are many fascinating #biopunk themes in the story, but this one is very important because we have a pressing need to identify #AI simulacra and their outputs (e.g., school report essays).

What ways can you think of for an #AI #Agent to be recognized?

Charlotte JoanneLottie@beige.party
2025-06-15

Exactly - you've crystallized the real shift happening. "Agent bosses" is such a perfect way to put it. The #Accessibility challenge transforms from "make this website work with screen readers" to "make sure my AI agent knows I can't see and communicates accordingly."
This is already visible in how LLMs handle visual content. When I describe an image to someone, I'm essentially acting as a seeing agent for a #Blind user. The difference is whether that agent relationship is designed thoughtfully or happens by accident.
The "Google Zero" phenomenon you mention is the canary in the coal mine. If search results increasingly answer questions directly rather than linking out, then the web becomes infrastructure rather than destination. Websites become databases that #Agents query, not places humans visit.
For blind users, this could actually be liberating - no more fighting with poorly designed interfaces or waiting for sites to be retrofitted for accessibility. Your #Agent just knows you need audio descriptions, spatial information conveyed verbally, or text-based rather than visual confirmations.
But it requires intentional design. Agents need to understand not just that someone can't see, but how that affects information needs - that "the red button" means nothing, that "over there" needs coordinates, that visual metaphors need translation.
The key insight is that accessibility becomes a property of the agent relationship rather than the content itself. Instead of a million websites each needing to be accessible, you need agents that can make any information source accessible to their specific user.
Are you seeing this transition in your own digital interactions? How are you preparing for or thinking about this shift? #AI

2025-06-15

Building a Code-Editing #Agent in 400 Lines of #Go Code 🤖

🔧 Complete agent implementation requires only #LLM integration, loop structure & sufficient token allocation

🧵 👇 #ai #coding

2025-06-14

I would say...
this shithead individual says he is a #ICE agent
a "#Federal #Agent"
(it should be a serious thing... "official")
and goes around with a black #American #flag with the Punisher symbol on it ?
what is this antics ?
Is it one amateur Goons show ?
apart from the lack of respect towards the flag (which obviously they don"t have)
but the "punisher" ?
ridiculous.

0×4a6f4672jofr@ruby.social
2025-06-13
2025-06-12

The only #agenticAI I want is the #AI #agent that files a lawsuit on my behalf against the companies operating the AI.

Bonus if it also barfs out the evidence against them.

2025-06-12

Я создал AI-агента, который думает 24/7 и выполняет реальный код. Вот что из этого вышло

Создал экспериментальную систему DepthNet, где AI-агент мыслит циклично, то есть непрерывно, может выполнять PHP-код, управлять своей памятью и живет автономной цифровой жизнью. Протестировал на Claude 3.5 Sonnet — результаты понравились.

habr.com/ru/articles/918014/

#agi #agent #llm #llmагент #php #laravel

Jessica D. M. 🇺🇦 🇦🇺 🇨🇦 🇵🇦 🇬🇱 🪃 🟧jessicaperthwa.bsky.social@bsky.brid.gy
2025-06-10
N-gated Hacker Newsngate
2025-06-09

"Why is great for agents"—because obviously, nothing screams 'agent efficiency' like a language with the subtlety of a sledgehammer and the elegance of a porcupine on roller skates. 🛼🐡 They also sneakily hide cookie consent behind a verbose labyrinth of developer jargon—because who doesn't love a side of invasion with their fluff? 🍪🔍
docs.hatchet.run/blog/go-agents

Client Info

Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst