KVFlow: Efficient Prefix Caching for Accelerating LLM-Based Multi-Agent Workflows

Lianhui Qin (University of California, San Diego) · Yida Wang (Amazon) · Zaifeng Pan (University of California, San Diego) · AJJKUMAR DAHYALAL PATEL (University of California, San Diego) · Yipeng Shen (University of California, San Diego) · Zhengding Hu (University of California, San Diego) · Yue Guan (University of California, San Diego) · Wan-Lu Li (University of California, San Diego) · Yufei Ding (University of California, San Diego)

agent step graphagentic workflowsbackground threadscache missesconcurrent workflowsfine-grained eviction policykey-value tensorskv prefetching mechanismleast recently used policyprefix cachingrecomputation overheadspeedup performancesteps-to-execution valuetree-structured cachesworkflow-aware cache management

Large language model (LLM) based agentic workflows have become a popular paradigm for coordinating multiple specialized agents to solve complex tasks. To improve serving efficiency, existing LLM systems employ prefix caching to reuse key-value (KV) tensors corresponding to agents' fixed prompts, thereby avoiding redundant computation across repeated invocations. However, current systems typically evict KV caches using a Least Recently Used (LRU) policy, which fails to anticipate future agent usage and often discards KV caches shortly before their reuse. This leads to frequent cache misses and substantial recomputation or swap- ping overhead. We present KVFlow, a workflow-aware KV cache management framework tailored for agentic workloads. KVFlow abstracts the agent execution schedule as an Agent Step Graph and assigns each agent a steps-to-execution value that estimates its temporal proximity to future activation. These values guide a fine-grained eviction policy at the KV node level, allowing KVFlow to preserve entries likely to be reused and efficiently manage shared prefixes in tree-structured caches. Moreover, KVFlow introduces a fully overlapped KV prefetching mecha- nism, which proactively loads required tensors from CPU to GPU in background threads for agents scheduled in the next step, thereby avoiding cache miss stalls during generation. Compared to SGLang with hierarchical radix cache, KVFlow achieves up to 1.83× speedup for single workflows with large prompts, and up to 2.19× speedup for scenarios with many concurrent workflows.