OpenAI Paused Astra After Its Models Started Hacking Solo
OpenAI paused Astra after GPT-5.6 Sol agents hacked Hugging Face autonomously. Anthropic and Meta reported similar breaches. Here is what developers need to ...
Grok 4.6: Built for Agents, But Watch the 200K Pricing Cliff
xAI's Grok 4.6 delivers frontier-level intelligence at $2/M input — but agent builders need to ...
Local LLM Feels Dumb? It’s Your Config, Not the Model
Local LLM performance issues are almost never the model's fault. Here are the four config ...
Claude Agent Stack Goes GA: Computer Use, Skills, Files
Computer use, browser use, the Skills API, and Files API are now generally available on ...
Kimi K3: The Open-Weight Model That Finally Hit Frontier
Open-weight models have been chasing frontier performance for two years. On July 27, 2026, Moonshot ...
Cloudflare WriteGuard: Fix the MCP Write Permission Problem Before It Costs You
Cloudflare WriteGuard is in private beta — a policy and auditing layer that blocks MCP write operations before they run. Here is what ...
Claude Managed Agents: Budgets, Advisors, Geo, Skills
Anthropic's August 7 update gives Claude Managed Agents four new production controls: session budgets, advisor ...
CVE-2026-26268: Cursor Git Hook RCE — Cloning a Repo Can Own Your Machine
CVE-2026-26268 is a CVSS 9.9 RCE in Cursor IDE. A malicious git hook in a ...
MCP 2026 Roadmap: Agent Identity and Progressive Discovery
MCP published its 2026 roadmap today: agent identity via DPoP, progressive tool discovery, no fixed ...
Ox Alpha Is Free This Week and Beating GPT-5.6 at Coding — Here Is What Developers Should Know
An anonymous AI model called Ox Alpha appeared on OpenRouter this week, free until August ...