SourcesIntentCUA Multi-Agent Desktop Automation 74.83% SuccessarXiv·high signalXBlueskyLinkedInCopy linkMulti-agent framework achieving 74.83% task success rate. Planner, Plan-Optimizer, Critic over shared memory. Cross-application skill transfer. Outperforms RL and trajectory baselines.SourceSource pagearXiv↳ Follow the threadPolicy dependency / Stack layerRIPPLE: an edit confined to one prompt-policy segment changes downstream behavior, so replay candidate edits after previously accepted ones before persistingarXiv 2609.12127Stack layer / ContrastReflexion-Style Verbal Memory Sometimes Lowers Success Versus Plain Retry, and Replay Experiments Show WhyarXiv 2609.12404Policy dependency / Stack layerSeqMoE Reaches 80.22% of Full-Load MoE Performance With Only 45% of Experts Resident in Device MemoryarXiv 2609.12978Stack layer / ContrastDifficulty-aware topology selection beats always-hierarchical multi-agent coding by 4.1 points at 40% of the costarXivStack layer / Update threadSplitting a CTF Task Into Isolated Sub-Contexts Lets a Local gemma-4 Solve 18.52% of Challenges Standard Agent Loops FailarXiv 2609.12839Stack layer / ContrastSplitting Decode by Attention Type Instead of by Operator Buys 31-56% More Tokens Per JoulearXiv 2609.13134Stack layer / Follow-up threadPlanting Benign-Sounding Reasoning in an Agent's Context Evades Chain-of-Thought Monitors 25-33% of the TimearXiv 2609.15989Stack layer / ContrastReconstructing Facts From Feed-Forward Residual Vectors Answers Single-Fact Questions in Two-Million-Token ContextsarXiv 2609.12686