Hacker News

New stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Linux Command Line Tricks that you do not know yet(lovethepenguin.com)
    discuss
  2. Show HN: Newdo – a todo list where the list is an output, not an input(beaverminds.com)
    discuss
  3. Grit your teeth and ship it(seangoedecke.com)
    discuss
  4. Discourse (open source forums) adds Discord-style voice and video calls(discourse.org)
    discuss
  5. Recursive self-improvement of AI research agents(arxiv.org)
    discuss
  6. Show HN: Analysis of Games from Chess Olympiad with Opus 5.5 and Stockfish(unremarkable.info)
    discuss
  7. Microsoft disrupts AI-assisted platform that compromised 12,000 accounts(arstechnica.com)
    discuss
  8. Why Tool AIs Want to Be Agent AIs(gwern.net)
    discuss
  9. Fearless SIMD v1.0 is here(linebender.org)
    discuss
  10. M6 Mac Mini uses soldered NAND storage again(macrumors.com)
    discuss
  11. Show HN: Yass – Yet Another Spec Syntax Enters Beta(github.com/shakefu)
    1comments
  12. Building a docs site that screenshots itself(semaloop.com)
    1comments
  13. There is no epidemic of loneliness, but there is an epidemic of scurvy(experimental-history.com)
    discuss
  14. Claude Code Hooks Explained: The Deterministic Layer Around Your Agent(blakecrosley.com)
    discuss
  15. Jevify skill – Gets your existing agents running on Jev
    discuss
  16. Show HN: use-citations – Claude Code skill for citation backed answers(github.com/sshah03)
    discuss
  17. Show HN: Daily-updated open dataset of flight fare history by route and month(github.com/deusacc)
    discuss
  18. CobaltC now has a compiler (for Linux)(github.com/strawberry9)
    discuss
  19. First Petabit-Class Transoceanic Subsea Cable(fb.com)
    discuss
  20. Ask HN: Engineers on highly polished products, how are you using agentic coding?
    1comments
  21. Which AI model should your team use for Ansible(steampunk.si)
    discuss
  22. Unreal Agent An async-first agent harness(github.com/unreallabsai)
    discuss
  23. Transformers now runs llama.cpp quants(huggingface.co)
    discuss
  24. Australia's giant data-centre boom(economist.com)
    1comments
  25. Avhash: BlurHash replacement, using a AV1 frame with container headers stripped(github.com/keithamus)
    discuss
  26. Show HN: Ceptile AI Search platform which combines web results with AI chat(ceptile.com)
    discuss
  27. Claude Opus 5.5 – Pelican Game(riba2534.cn)
    2comments
  28. I Gave My Life over to Meta's A.I. Agent and Was Blown Away(nytimes.com)
    1comments
  29. El Niño projected to cause 450k deaths in next six months(theguardian.com)
    1comments
  30. Mastering LLM Inference Optimization(machinelearningmastery.com)
    discuss

Show HN: use-citations – Claude Code skill for citation backed answers

1 pointsby 41m agogithub.com
0 comments
This is a Claude Code skill for research questions in law, tax, medicine and similar fields, or for anyone wanting to ask knowledge-based or research questions while in a Claude Code shell really. It gets claude to save the source documents, answers from them, and builds an artifact/report where each sourced phrase opens in a slide out drawer with the relevant passage highlighted in the full source doc (quick video demo of the artifact is in the linked repo).

Tools like Harvey, Legora, and OpenEvidence provide a lot more value than just this, so it isn’t a replacement. I imagine, though, that there are some people in those fields around the world who might have a Claude/GPT subscription and not one of the domain specific AI tools. I know CLI might not be natural to all in the those industries, but I’ve found Claude Code works better for me as a harness for this kind of research than Chat, and this adds the citation and grounding pieces, to some degree, that those tools provide.

It is MIT licensed. Suggestions and pull requests are welcome. If you’re at a firm with Claude/GPT subscriptions and want help getting people set up with this, feel free to ask.

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