Hacker News

Top stories

Live mirror
30 storiesupdated just nowView source snapshot
  1. Human brain is two separate organs, Stanford Medicine-led research finds(stanford.edu ↗)
    32comments
  2. GPT-6 Astra Solves a WWI German Radio Cipher(prinzai.com ↗)
    8comments
  3. If math is more than proof, we need to better celebrate the rest of it(terrytao.wordpress.com ↗)
    10comments
  4. San Francisco Onion Futures Company(onionfutures.com ↗)
    55comments
  5. Android 17 is the first since 3.x to add new APIs without releasing to the AOSP(grapheneos.social ↗)
    371comments
  6. Typesafe-computer-use drives a Mac toward a goal for 1/50th of a cent per step(github.com/awlevin ↗)
    25comments
  7. Science Is Open Software(jepedersen.dk ↗)
    27comments
  8. SDCC – Small Device C Compiler(sourceforge.net ↗)
    15comments
  9. Cloudflare Quick Tunnels(cloudflare.com ↗)
    275comments
  10. NASA-IBM Lunar Foundation open-Source Geospatial AI Model(usra.edu ↗)
    1comments
  11. Why building a Rust LSP is hard(rust-glancer.github.io ↗)
    26comments
  12. You can run Git on object storage if you re-make packfiles(tigrisdata.com ↗)
    6comments
  13. Saving another 100TB of RAM(cloudflare.com ↗)
    65comments
  14. How to Write with an LLM(sockpuppet.org ↗)
    316comments
  15. Ctenophores: Wonders of Biology(quantamagazine.org ↗)
    4comments
  16. How OpenAI Used Its Own LLMs to Design Its Jalapeño Chip(ieee.org ↗)
    78comments
  17. The first new cat species discovered in 100 years(nationalgeographic.com ↗)
    94comments
  18. Goroutine Leak Profiles(go.dev ↗)
    2comments
  19. OpenJev(openjev.com ↗)
    257comments
  20. Show HN: Cactus Needle 3: 8-29MB automation models can match DeepSeek V4 Flash(cactuscompute.com ↗)
    88comments
  21. Photon-Emission-Guided Laser Fault Injection Enables RP2350 Secure Debug(ledger.com ↗)
    66comments
  22. Veronese's Dogs(publicdomainreview.org ↗)
    discuss
  23. Cache-to-Cache: Direct Semantic Communication Between LLMs (2025)(arxiv.org ↗)
    12comments
  24. The Farnese letter(simonklee.dk ↗)
    6comments
  25. Minimal Phone 2(minimalcompany.com ↗)
    214comments
  26. Stepfun Step 5 Preview (LLM): On AA Pareto frontier(artificialanalysis.ai ↗)
    1comments
  27. Xcode 27.1 Beta Release Notes(developer.apple.com ↗)
    97comments
  28. Cyclomatic Complexity in C#(ndepend.com ↗)
    18comments
  29. Inside ZCode: Silently uploading your Git history to the cloud(ferstar.org ↗)
    97comments
  30. Warez: The Infrastructure and Aesthetics of Piracy (2021)(archive.org ↗)
    52comments

Anthropic to limit Using third-party harnesses with Claude subscriptions

20 pointsby 5mo ago
8 comments
Starting April 4 at 12pm PT / 8pm BST, you’ll no longer be able to use your Claude subscription limits for third-party harnesses including OpenClaw. Instead, they’ll require extra usage.

Your subscription usage still covers all Claude products, including Claude Code and Claude Cowork. To keep using third-party harnesses with your Claude plan, enable extra usage for your account. This will be enforced April 4 starting with OpenClaw, but this policy applies to all third-party harnesses and will be rolled out to more shortly (read more).

To make the transition easier, we’re offering a one-time credit for extra usage equal to $200 on your Team plan. Redeem your credit by April 17. We’re also introducing discounts when you pre-purchase bundles of extra usage (up to 30%) for your team.

We’ve been working to manage demand across the board, but these tools put an outsized strain on our systems. Capacity is a resource we manage carefully and we need to prioritize our customers using our core products. We appreciate your support as we make these changes.

5mo agoHN ↗

I feel they should leave an opening for claws that use the Claude code sdk (like nanoclaw) because they will still operate on behalf of the main user. The same rate limits can apply as for CC, so why not?

Or even let us maybe use haiku only with claws?

But if this becomes a hassle, I won’t mind giving my $200/month to OpenAI instead.

Cuz for whatever reason, they seem to have way higher quotas even in the $20 plan.

Time for nanoclaw to add an adapter to work with other SDKs.

5mo agoHN ↗

Are there any better suggestions or better ways?

5mo agoHN ↗

API direct. $20/mo sub gives you the model access, then pay per-token for what you actually use through your own setup. You lose the convenience of a UI, but you keep control over how and where you run your prompts. For code-heavy workflows specifically, Cursor is unaffected since they have their own deal. The real hit is on general-purpose harnesses like OpenClaw.

5mo agoHN ↗

Could've allowed us to use Haiku, Anthropic be reducing cc limits the same speed they're launching features.

5mo agoHN ↗

There’s quite a large misconception that third party harnesses are banned. Anthropic specifically said the agent sdk is still allowed, as long as you build on top of it - you can create really powerful software that drives Claude code in an efficient, performant way other than the raw Claude code terminal app https://x.com/bcherny/status/2040207998807908432?s=48