Loading...
The August 2025 Nx s1ngularity attack turned installed AI coding agents into credential scanners. 2,349 secrets, 1,079 repos compromised. The --dangerously-skip-permissions flag was the entire attack. Here's the architectural fix.
A Chinese team called EvoMap matched OpenAI Codex on six multi-agent benchmarks while running 3x cheaper — by throwing out the planner agent entirely. Codex bets on one super-individual; EvoX bets on a deterministic swarm. The bottleneck was never the model.
NVIDIA's open models are a distribution layer for its hardware and software ecosystem: shared foundations, open recipes, and production deployment turn free weights into infrastructure adoption.
Anthropic's Claude Code team revealed they cut more than 80% of the system prompt when targeting Opus 5 and Fable 5 — coding evals did not regress. Here is the six-rule Then/Now shift, and the deletion checklist for your own CLAUDE.md.
Claude Opus 5 turns inference policy into a production concern. A task-aware framework for effort, tools, retries, verification, fallback, latency, and cost.
Cursor's internal Agent swarm ran the same SQLite rewrite twice and got 70,000 conflicts the first time, 47 the second. The hardware and the models did not change. The org chart did. Here is what every team shipping coding Agents should steal on Monday.
DeepSeek just shipped V4-Pro (1.6T/49B active) and V4-Flash (284B/13B active) — both with 1M-token context, MIT license, and pricing that undercuts every frontier lab. The architecture (CSA + HCA attention) cuts KV cache 8.7× vs V3.2. Near-Opus quality, 30% of the price.
On July 21, 2026, JetBrains shipped a repository intelligence layer that reduces Claude Code, Codex CLI, and Junie turns by up to 68%. It signals that the coding-agent bottleneck has moved from models to codebase indexing.
Cloudflare is bringing shared dictionary compression (RFC 9842) to production on April 30 — a delta-compression technique that turns a 272KB JS bundle into 2.6KB after a one-line change. After 18 years of failed attempts (SDCH in 2008, abandoned by Chrome in 2017), the new standard finally fixes the security and CORS issues that killed the first attempt. Here's how it works, why it took so long, and what changes for high-deploy web apps.