Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
The expert pushback challenges the factual basis of the Commerce Department's export controls, with Moussouris arguing the research cited by the administration demonstrates defensive security capabilities rather than a genuine bypass of Fable 5's safeguards.
The extension closes the official Claude Code IDE integration gap for Visual Studio, bringing native diff-based code review controls to a platform that previously had no supported path beyond running Claude in a terminal.
The pattern directly addresses token waste and rule conflicts in Claude Code projects by replacing a single always-loaded context file with scoped imports, so each session carries only the rules relevant to the task at hand.
The tool fills a gap left by AI coding assistants that either omit usage data or silo it per-tool, giving developers a single aggregated cost and token view across Claude Code, Codex, and Cursor without requiring any changes to those tools.
The work shows that real hardware feedback is the critical missing ingredient for LLM agents to autonomously replace expert-driven MCU optimization, turning a previously manual, multidimensional process into a closed-loop pipeline that outperforms human experts within seven iterations.
An open-source coding agent from Xiaomi claiming to outperform Claude Code on long-horizon tasks is a notable development in the agentic coding tooling space.
Linksee's `PreToolUse` gate introduces a mechanism that can actively block AI agent actions that contradict declared product intent, moving drift detection from a passive warning into an enforcement layer.
AgentFairBench shifts fairness measurement from LLM text outputs to agent decisions in consequential domains, and its arity-matched null methodology corrects a ~2.4× overstatement of disparity that prior comparison approaches produce.
RetailBench exposes that current LLMs cannot sustain coherent long-horizon decision-making in economically grounded environments, with most models failing to complete even a 180-day simulation and all falling substantially short of an oracle policy on net worth and sales.
Open-SWE-Traces provides a large-scale, permissively licensed, multilingual trajectory dataset that enables fine-tuning of open-source LLMs for autonomous software engineering — directly addressing the data scarcity the paper identifies as the primary bottleneck on this path.