Tangram: non-uniform KV cache for cheaper multi-turn serving.

Shared entity: Directly / Same source domain / Shared topic / Earlier coverage / Tension

SplitAgent: Privacy-Preserving Enterprise Agents

Both cover Directly; reported by the same outlet (arxiv.org); overlapping topics (agent, directly).

AgentSentry: Defending Against Indirect Prompt Injection in Tool-Using Agents

Both cover Directly; reported by the same outlet (arxiv.org); overlapping topics (agent, directly).

Shared entity: Directly / Same source domain / Shared topic / Earlier coverage

Vortex makes sparse attention programmable for long-generation agents.

Both cover Directly; reported by the same outlet (arxiv.org); overlapping topics (agent, directly, serving).

Memory-R1: RL-Based Agent Memory Management

Both cover Directly; reported by the same outlet (arxiv.org); overlapping topics (agent, directly, memory).

Learning to Share: Selective Memory for Parallel Agent Teams

Both cover Directly; reported by the same outlet (arxiv.org); overlapping topics (agent, directly, memory).

Context Pollution: LLMs Don't Benefit From Their Own Words

Both cover Directly; reported by the same outlet (arxiv.org); overlapping topics (agent, directly, multi-turn).

Shared entities / Same source domain / Earlier coverage

The 1/W Law: Routing Topology Beats Hardware Upgrades for Inference Efficiency

Both cover Directly, GPU; reported by the same outlet (arxiv.org); earlier Directly coverage from 2026-03-21.

Shared entity: Multi / Same source domain / Shared topic / Earlier coverage

CachePrune: Multi-Tenant LLM Serving Has a Side-Channel Problem.

Both cover Multi; reported by the same outlet (arxiv.org); overlapping topics (cache, serving).