today's signal · no scroll
the brief
Today felt like stack-shaping infrastructure and agent tooling day: OpenAI dropped the MRC networking protocol for AI clusters while Anthropic expanded on multiple fronts with AWS GA, an Agent View, and higher limits. LangChain shipped a full agent stack, LiteLLM open-sourced safer agent sandboxes, and Codex added background Chrome control—all against the backdrop of urgent RSC/Next.js vulns and falling voice API prices.
Multiple vulnerabilities affecting RSC and Next.js were published; Cloudflare WAF mitigates disclosed DoS vectors, but developers should update dependencies immediately to patch upstream.
Multiple security vulnerabilities affecting React Server Components and Next.js have been disclosed. We strongly recommend updating your applications immediately. Cloudflare WAF managed rules already mitigate the disclosed denial-of-service vulnerabilities, and we are
New open Multipath Reliable Connection protocol to run large AI training clusters faster and more reliably, reducing GPU idle with backing from AMD, NVIDIA, Intel, Microsoft, Broadcom.
We’ve partnered with @AMD, @Broadcom, @Intel, @Microsoft, and @NVIDIA, to release Multipath Reliable Connection (MRC), a new open networking protocol that helps large AI training clusters run faster and more reliably, with less wasted GPU time. openai.com/index/mrc-supe…
AWS customers can access full Claude API features with AWS authentication, billing, and commitment retirement, streamlining procurement and deployment for enterprise workloads.
The Claude Platform on AWS is now generally available. AWS customers get the full set of Claude API features, with AWS authentication, billing, and commitment retirement. pic.x.com/9WPrEgCpkg
Research preview centralizes all agent sessions into one list, improving multi-session management and continuity for iterative coding workflows.
New in Claude Code: agent view. One list of all your sessions, available today as a research preview. pic.x.com/NnbsAQjSPW
Doubling 5‑hour limits for Pro, Max, and Teams, removing peak-hour reductions, and substantially increasing Opus API rate limits to unblock heavier coding and integration use.
Usage limits are up, effective today we're: 1) Doubling Claude Code's 5-hour limits for Pro, Max, Team and seat-based Enterprise plans 2) Removing peak hours limit reduction on Claude Code for Pro and Max plans 3) Substantially raising our API rate limits for Opus models x.com/claudeai/statu…
New LangSmith Engine, SmithDB, Sandboxes, Managed Deep Agents, LLM Gateway, Context Hub, and Deep Agents 0.6 target faster end‑to‑end agent development and operations.
We just shipped tons of new products to accelerate the full agent development lifecycle: langchain.com/blog TLDR: ✅ LangSmith Engine ✅ SmithDB ✅ Sandboxes ✅ Managed Deep Agents ✅ LLM Gateway ✅ Context Hub ✅ Deep Agents 0.6 pic.x.com/BOpHFBTVpd
Self‑serve pricing drops up to 55% for Text‑to‑Speech, 45% for Speech‑to‑Text, and 20% for Agents, making production voice features more accessible.
We’ve cut ElevenAPI and ElevenAgents pricing for self-serve developers. From today, prices are lower and you can pay as you go: - Text to Speech is now up to 55% lower cost - Speech to Text is now up to 45% lower cost - Agents is now up to 20% lower cost Performance, quality, pic.x.com/yCJnz9mObP
Run Claude Code, Codex, Hermes and more in isolated Kubernetes sandboxes where agents never see your real API keys, improving security for multi‑model deployments.
We're open sourcing LiteLLM Agent Platform Run Claude Code, Codex, Hermes or any coding agent in isolated K8s sandboxes The agent never sees your real API keys. Here's how 👇 pic.x.com/NsuT1ljpNQ
Codex can now control Chrome tabs headlessly, enabling background browsing and automation for agent workflows without fragile foreground focus hacks.
Codex can now drive Chrome tabs in the background: x.com/jameszmsun/sta…
New arXiv work uses Bradley–Terry aggregation across parallel reasoning paths, suggesting a reproducible route to faster multi‑trajectory thinking beyond linear token chains.
Here are the strongest technical items I could verify from the last 24 hours in the AI engineering / research feeds you shared, with canonical URLs: 1. **OpenDeepThink: Parallel Reasoning via Bradley–Terry Aggregation** — a fresh arXiv cs.AI paper on parallel reasoning and aggregation. Canonical URL: https://arxiv.org/abs/2605.15177 [3] 2. **Dual-Dimensional Consistency: Balancing Budget and Qua — citation: https://arxiv.org/list/cs.AI/new
Matt Pocock breaks down Anthropic’s “dedicated monthly credit,” arguing it effectively boosts AFK usage 5–20× and clarifies how quotas change real developer workflows.
Anthropic has given us a "dedicated monthly credit" Which, in effect, slashes AFK usage limits of Claude Code by ~5-20X Here's how it affects you: pic.x.com/1nCCheqEC4
Jason from the Codex team shares a practical draft of daily primitives for more reliable, capable agents—an invitation for feedback from builders pushing Codex in production.
jason from the codex team here, heres a draft on codex maxxing and the primatives i use on a daily basis jxnl.github.io/blog/writing/2… would love any feedback