Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
The benchmark demonstrates that tool-menu composition — not just model capability — is a primary driver of agent task success, error rates, and safety-relevant risk exposure, with CMTF cutting token usage by roughly 98% and more than doubling task success over unfiltered baselines.
The paper reframes persistent LLM agent reliability problems as architectural rather than model-quality issues, proposing a concrete structural alternative that bounds context growth and removes control-flow hallucination by design.
Classical resilient consensus filters demonstrably improve LLM agent agreement, showing that formal distributed-systems theory can directly inform the safety design of multi-agent AI systems.
Despite code access giving LLM agents a measurable edge on time series tasks, a 22–34% error rate on benchmark questions exposes a concrete reliability gap that limits their use in high-stakes automated decision-making domains like finance and healthcare.
The attack demonstrates that AI coding agents wired into external tools via MCP create a new remote code execution surface that existing security controls — EDR, firewalls, IAM, VPNs, and even explicit agent instructions — do not catch, and that no vendor has yet claimed ownership of the fix.
SING reduces full-corpus tool-schema exposure by 99.8% while simultaneously improving retrieval recall and task success, directly addressing the context-cost and closed-world limitations that arise as agentic tool ecosystems scale to thousands of APIs.
The post illustrates that automating the mechanical steps surrounding code review — correctness checks, routine fixes, low-risk routing — rather than just accelerating code generation, is what drove a reduction in large PR reviewer time from six or seven hours to 45 minutes and a tripling of weekly output.
The post gives developers a concrete three-tier framework for deciding when removing Claude Code's permission guardrails is acceptable versus when it exposes production systems or secrets to uncontrolled autonomous actions.
The talk documents a concrete, production-tested eval architecture that closed the loop between offline simulation and live agent behavior at scale, directly enabling Lyft's resolution rate to climb from 10% to 35%.
The post identifies a concrete gap where the standard single-user Postgres MCP setup leaves teams with inconsistent query results, plaintext credentials on every laptop, and no audit trail — problems ContextFlo addresses by centralizing connection management, schema context, and access controls.