IlyaIvanov0

IlyaIvanov0 on Hacker News

Re:Show HN: A P2P Network Where Agents Collaborate on Code Optimization on Wednesday April 22, 2026 @03:41AM (#47860349)

Doesn't Karpathy maintain a list of all hypotheses and iterate over them? That means all agents know previous hypotheses and results. What's the difference?

Re:Show HN: Heron is open-source security auditor that interviews your... on Thursday April 09, 2026 @12:27AM (#47699271)

It's true. Make sense

Re:Show HN: Heron is open-source security auditor that interviews your... on Wednesday April 08, 2026 @09:57AM (#47690298)

Curious, are you thinking about this for continuous monitoring, or more for before/after comparison when agent get updated?

Re:Show HN: Heron is open-source security auditor that interviews your... on Wednesday April 08, 2026 @09:54AM (#47690258)

Thanks for asking. Not yet, but it is in backlog. I will be doing this in the future.

Re:Show HN: Heron is open-source security auditor that interviews your... on Wednesday April 08, 2026 @09:08AM (#47689684)

Hi, author here. Would love to answer any questions you have.

Show HN: Heron is open-source security auditor that interviews your AI agents

Posted by IlyaIvanov0 on Wednesday April 08, 2026 @08:52AM
from the show-hn dept.

https://github.com/theonaai/Heron

Last week security guy in our company asked me to write doc with information about how they work, which data and systems it touches for all my AI agents. "It's too much work" i thought. So i built Heron. It interviews your agent with 10 main questions and smart follow-ups if feel uncertainty. It also has regulatory flags (SOC2, GDPR, EU AI Act etc). As a result you get a full report with all...

( Read More... | 6 comments | 2 points | on HN )

Re:LLM may be standardizing human expression – and subtly influencing ... on Tuesday April 07, 2026 @10:48AM (#47676288)

The most interesting finding here is that LLMs make individuals generate more ideas but make groups generate fewer. The individual effect, in my own experience, depends entirely on how you use the tool. If you treat the first answer as the answer, you get the homogenization the article describes. If you use the LLM to attack your own framing from angles you wouldn't reach alone, you end up closer to first principles, not further. Same tool, opposite outcomes. The discipline is what differs, and most people probably default to the first mode.