Anthropic’s September threat report details eight months of disrupted Claude misuse, from a Russia-aligned espionage campaign and AI-run exploit foundries to Chinese labs accused of industrial-scale distillation. It says most of the serious cases used multi-agent frameworks with humans mainly setting targets and reviewing loot.
Released Oct 25, an opt in ENABLE SKILL LAZY LOADING setting lists selected skills for the model to load only when needed instead of sending every skill's instructions each message, and MCP connections now work with stateless 2026 07 28 servers. It also carries security fixes the team will detail after CVE assignment, so update soon, and note that saved tool server and Open Terminal connections now stop working unless an admin turns on ENABLE DIRECT INTEGRATIONS…
Released Oct 24, d3 brings vLLM Semantic Router's 2B to 27B decision models to Ollama: pass text, structured state or an image and get a classification, yes/no answer or probability scored options, e.g. ollama run d3 "..." ask "Is this a billing issue?" . It's a cheap local gate for routing or approving an agent's next step…
The Apache 2.0 InnovationHub project wraps the pi coding agent in Monaco tabs, terminals and diffs, runs parallel agents in separate worktrees, and keeps project intent in a graph of SPEC.md files the agent reads and updates. It's an experimental incubator project; install with the one line script in the repo and run thinkrail (TechNewsReel)…
Corgi runs multiple coding agent sessions side by side with a live activity feed, a jump to next unread agent shortcut, a unified launcher with session search and usage tracking, and image paste that works over SSH. The solo developer says multi machine remoting is next; Windows works via WSL2…
The 33B dense model from BOSS Zhipin's Nanbeige lab self reports 80.9 on Terminal Bench 2.1 and 65.5 on DeepSWE versus 73.0 and 42.5 for Qwen3.8 27B, though Qwen still edges it on SWE Bench Pro (61.7 to 60.3). It runs on vLLM or SGLang with Qwen3 tool call parsing, and the scores are the lab's own, so test it on your repos before swapping your local coding agent (MarkTechPost)…
Spotted something we missed? Start a thread.
