Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Proof that computers can't do everything(youtube.com ↗)
    discuss
  2. Lithography Machines (Proposal)(drive.google.com ↗)
    discuss
  3. Turn any Llama-server into a jev system one endpoint(github.com/khimaros ↗)
    discuss
  4. GSoC 2026: Port the Enlightenment desktop environment to NetBSD, part 2(netbsd.org ↗)
    discuss
  5. TXR: An Original, New Programming Language for Convenient Data Munging(nongnu.org ↗)
    discuss
  6. Meta's Muse Is Better at Surveilling Than Helping Me(wired.com ↗)
    discuss
  7. OSS, security and Funding: libjpeg-turbo
    2comments
  8. Grit your teeth and ship it(seangoedecke.com ↗)
    1comments
  9. GoldenEye 007 Native PC Port(jkdansereau.github.io ↗)
    discuss
  10. Covert Caches: When Is a Cache, a Cache?(parallelprogrammer.substack.com ↗)
    discuss
  11. Mini-AGI – dynamic continual learning model trained from scratch on 8GB VRAM(github.com/volotat ↗)
    1comments
  12. A build graph that rolls dice(fzakaria.com ↗)
    discuss
  13. Hawking by Graham Farmelo review – the dark side of genius(theguardian.com ↗)
    discuss
  14. Investment Seer: Agentic AI-Powered Stock and Crypto Analyst(investmentseer.com ↗)
    discuss
  15. Ant Mill(wikipedia.org ↗)
    discuss
  16. Heretic removes restrictions from language models(heretic-project.org ↗)
    1comments
  17. Meslo-Font: Customized Version of Apple's Menlo Font(github.com/andreberg ↗)
    discuss
  18. The Docker API Ceiling: Why Portainer 3.0's New Capabilities Target Kubernetes(portainer.io ↗)
    discuss
  19. AI chatbots give wrong answers to financial queries 'most of the time'(ft.com ↗)
    2comments
  20. Ask HN: Is provider KV caching sufficient for agent swarms and long run agents?
    discuss
  21. Knowledge Refresh for Production RAG
    discuss
  22. Why Do We Need Human Mathematicians Anymore?(poshenloh.com ↗)
    1comments
  23. RBA Faces Key Rate Decision as Inflation Pressure Remains(coinmarketcap.com ↗)
    1comments
  24. Faked Images of Neurons in Major Journals(youtube.com ↗)
    discuss
  25. What do you think of a Java-like language that compiles to native code?(github.com/ironwood-lang ↗)
    2comments
  26. College Is Coming Apart After Successive, Overlapping Crises(theatlantic.com ↗)
    1comments
  27. The positive effect of walking on creative thinking(nih.gov ↗)
    discuss
  28. The president wanted to unleash AI. Then came Anthropic(politico.com ↗)
    discuss
  29. Show HN: Mahabharata Characters – 307 figures with gunas and decision primitives(huggingface.co ↗)
    1comments
  30. Show HN: Less Prompts, More Guardrails(yasyf.com ↗)
    discuss

Ask HN: Is provider KV caching sufficient for agent swarms and long run agents?

1 pointsby 54m ago
0 comments
I’m trying to build a side project in the inference space. I’ve been talking to a few inference engineers and startups and I’ve been hearing how annoying it is to not have manual control over the KV cache at times and just constantly being subject to the black box caching methods of their inference providers. It is particularly annoying for agent swarms when you want to fork agents from the same cached prefix or manually store a cache for a longer period for a future agent to hit later.

Im curious if this problem is consistent across multiple people and if there are any solutions for it that people know about.

A quiet thread, for now.Start the conversation on HN ↗