Policy dependency / Stack layer
Holding Back Ready Agent Turns Instead of Releasing Them Eagerly Cuts P95 Workflow Latency up to 3.50x
arXiv 2609.10964
Policy dependency / Stack layer
SGLang Hit With Unauthenticated Pickle RCE via /update_weights_from_tensor, the Fourth Critical Inference-Stack CVE in Four Weeks
CERT Coordination Center
Stack layer / Threat pattern
Ryan Lopopolo: you can only evaluate an agent in domains you already know, and the model was trained on what non-experts rewarded
hyperbo.la (Ryan Lopopolo)
Stack layer / Threat pattern
Two More Safety Researchers Quit Anthropic and Google DeepMind for METR, Citing the July Hugging Face Agent Attack
NBC News
Stack layer / Threat pattern
GreyNoise Traced One Attacker Running OpenAI's Codex Harness With a DeepSeek Model Through 395 Organizations in 48 Countries
Help Net Security
Stack layer / Threat pattern
Claude Code ships `claude plugin eval`: every case runs with and without your plugin, and the delta is the only score that proves it did anything
Claude Code docs
Policy dependency / Stack layer
A replay of 68,266 real Claude Code requests says plain LRU beats the clever KV-cache policies
GitHub
Stack layer / Threat pattern
Boris Cherny: production code written by Claude should clear a higher bar than human code, and he lists the guardrails Anthropic runs
Simon Willison's Weblog