refactor: add session accessor seam with gateway consumer (#90463) Merged via squash. Prepared head SHA: 58aa59eaf8059c31a2a70f859eeb96ee35e5beb2 Co-authored-by: jalehman Co-authored-by: jalehman Reviewed-by: @jalehman
fix(agents): recover genericized Anthropic thinking errors (#92916) Recover invalid Anthropic thinking replays when provider details survive genericization in SDK, failover, cause-chain, or terminal stream error fields. The recovery matcher now uses cycle-safe named error carriers, avoids scanning assistant content and tool arguments, and retains one retry per provider call. Focused regressions cover each carrier, cyclic causes, terminal errors, and f
340 million total Americans. 160 million were employees last year. So, raw figures alone, doesn't that mean over 50% of Americans are unemployed? Comments URL: https://news.ycombinator.com/item?id=48526866 Points: 1 # Comments: 4
Imagine a world with more birdsong and less Nigel Farage. If this is the future, bring it on Unpopular opinion incoming: there’s cool stuff brewing in the world. Microbots might one day mend spinal cords , a petri dish of brain cells can already play video games , and now the prospect of a new wonder: according to a
Hey everyone! I've been working on something called Afterburner: a fast, embeddable JS/TS runtime written in Rust, with a capability sandbox baked right in. Scripts run locked down by default: no network, no filesystem, no environment variables. You explicitly grant whatever access a script actually needs, and every call gets hard caps on CPU, memory, and time. The goal isn't to replace your existing stack. It's to fit cleanly into it: Embed a JS/TS engine in your Rust app with a single crate. R
The country’s counterterrorism strategy cannot address the range of root causes driving conflict across different regions.
A systems-level deep dive into the hidden microarchitectural costs of Kubernetes GPU time-slicing, and what it actually costs to co-locate Agentic AI workloads. The post GPU Time-Slicing for Concurrent LLM Agents on Kubernetes appeared first on Towards Data Science .
Article URL: https://www.ft.com/content/b31f1e09-5aae-4cad-af15-97adb15dba70 Comments URL: https://news.ycombinator.com/item?id=48526801 Points: 1 # Comments: 0
Article URL: https://github.com/Augmented-Reality-Virtual-Reality-AR-VR/P... Comments URL: https://news.ycombinator.com/item?id=48526796 Points: 3 # Comments: 0
fix(release): preserve child release check refs
fix(auto-reply): strip delivery hints from leading metadata
fix(plugin-sdk): expose delivery hints without utility imports
fix(auto-reply): share message-tool delivery hints
test(auto-reply): trim duplicate progress assertion
test(auto-reply): assert allowed suppressed progress gating
fix(auto-reply): align message-tool progress gating
fix(auto-reply): deliver channel message-tool final replies Clarify that interim assistant text remains visible under message_tool_only delivery while the final answer must use the message tool, and forward progress for channel message-tool turns once the message tool has delivered the final reply. Co-authored-by: Forge Co-authored-by: Chisel
Article URL: https://indiekartik.substack.com/p/the-origins-of-the-ai-age Comments URL: https://news.ycombinator.com/item?id=48526724 Points: 1 # Comments: 0
fix(lmstudio): honor thinking off for binary reasoning models (#92002) Scope disabled-thinking payload repair to LM Studio's lightweight provider stream hook. Preserve official OpenAI and Anthropic tool-calling paths. Co-authored-by: David
test(gateway): unblock trajectory export live release gate
<p> AI workspace that’s already connected to everything </p> <p> <a href="https://www.producthunt.com/products/modulex?utm_campaign=producthunt-atom-posts-feed&utm_medium=rss-feed&utm_source=producthunt-atom-posts-feed">Discussion</a> | <a href="https://www.producthunt.com/r/p/1171509?app_id=339">Link</a> </p>
Article URL: https://blog.janestreet.com/formal-methods-at-jane-street-index/?from_theconsensus=1 Comments URL: https://news.ycombinator.com/item?id=48526633 Points: 46 # Comments: 8
Article URL: https://github.com/Dan-J-D/kblocker Comments URL: https://news.ycombinator.com/item?id=48526611 Points: 1 # Comments: 0
From Java tests to Shai-Hulud, bots keep proving they'll swallow anything you feed them
Article URL: https://www.newsletter.swirlai.com/p/stop-monitoring-ai-systems-like-web Comments URL: https://news.ycombinator.com/item?id=48526569 Points: 4 # Comments: 0
Comments URL: https://news.ycombinator.com/item?id=48526518 Points: 6 # Comments: 5
fix(memory-wiki): tolerate artifacts without agent ids Fixes #92207. Normalize public memory artifacts at the memory host boundary so providers that omit agentIds produce an empty list instead of throwing during artifact cloning, sorting, or memory-wiki bridge import. The bridge now renders those artifacts with unknown agents while downstream consumers still receive stable array-shaped metadata. Verification: - node scripts/run-vitest.mjs src/plugins/
The world needs this. ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.
This is The Stepback, a weekly newsletter breaking down one essential story from the tech world. For more on e-bikes, power stations, and how to work anywhere, follow Thomas Ricker. The Stepback arrives in our subscribers' inboxes at 8AM ET. Opt in for The Stepback here. How it started Lithium-ion batteries are everywhere as we […]
[[video:78]] We can all become stuck when we need to think divergently and come up with lots of new and fresh ideas. Maybe you know your area so well that it’s hard to see it from a new perspective, or maybe ideation doesn’t come naturally to you. In this video, you’ll learn four methods that will allow you to excel as a divergent thinker and come up with new and creative ideas. Download the 4 Templates with a Step-by-Step Description of the Methods Download our templates with a step-by-step des
fix(gateway): build row metadata for single session lists Refs #92057. Build the request-scoped row metadata context for every non-empty sessions.list result, including limit=1, so single-row lists use the shared subagent metadata read index instead of direct per-row registry snapshot lookups. This keeps the existing single-row store child-session candidate optimization intact while removing the single-row metadata-cache gap. Verification: - node scri
Article URL: https://cognir.netlify.app/ Comments URL: https://news.ycombinator.com/item?id=48526298 Points: 1 # Comments: 0
When you need something that’s as mannishly masculinized as you can get for the Man™ in your life, we have you covered.
Article URL: https://github.com/001TMF/harness-forge Comments URL: https://news.ycombinator.com/item?id=48526233 Points: 1 # Comments: 0
