https://subagentic.ai/posts/cursor-marketplace-claude-code-backdoor/subagentic.aien2026-10-31T15:11:36.074ZCursor says tampered Claude Code builds sat in its marketplace https://subagentic.ai/posts/pydantic-ai-2-64-0-fallback/subagentic.aien2026-10-31T15:09:12.764ZBoxdSandbox runs Pydantic agent tools in an isolated cloud machine https://subagentic.ai/posts/claude-code-2-1-316-controls/subagentic.aien2026-10-31T15:07:49.640ZClaude Code 2.1.316 adds shared budgets and a subagent effort setting https://subagentic.ai/posts/opencode-2-2-4-persistent-terminals/subagentic.aien2026-10-31T15:07:26.119ZOpenCode 2.2.4 keeps desktop terminals across restarts https://subagentic.ai/posts/agent-framework-python-1-24-0/subagentic.aien2026-10-31T15:07:05.906ZAgent Framework 1.24.0 lets agents compact context mid-run https://subagentic.ai/posts/vercel-labs-agent-desktop/subagentic.aien2026-10-31T03:11:33.226ZVercel Labs ships agent-desktop for coding agents https://subagentic.ai/posts/aura-autopilot-open-source/subagentic.aien2026-10-31T03:08:41.976ZRamen open-sources Aura Autopilot for in-editor game testing https://subagentic.ai/posts/mistral-large-4-open-weights/subagentic.aien2026-10-30T15:11:10.847ZMistral releases the weights of improved Large 4 https://subagentic.ai/posts/openai-frontier-misalignment-monitoring/subagentic.aien2026-10-30T15:08:52.993ZOpenAI details a four-layer monitor for frontier agents https://subagentic.ai/posts/android-studio-bring-your-own-agent/subagentic.aien2026-10-30T15:08:11.501ZAndroid Studio Rabbit 2 ships Bring Your Own Agent https://subagentic.ai/posts/claude-self-assessment-misuse-flags/subagentic.aien2026-10-30T03:10:21.509ZAnthropic says Claude has been grading its own chats for misuse since June https://subagentic.ai/posts/gemini-4-carbon-staged-cyber-access/subagentic.aien2026-10-30T03:09:59.740ZGemini 4 Carbon opens to cyber defenders first, not as a general release https://subagentic.ai/posts/claude-code-2-1-314-write-boundary/subagentic.aien2026-10-30T03:07:10.798ZClaude Code 2.1.314 can refuse writes outside the working directories https://subagentic.ai/posts/openai-model-spec-five-values/subagentic.aien2026-10-30T03:06:48.585ZOpenAI reorganizes the Model Spec around five values and renames the model GPT