ClawVM: Harness-Managed Virtual Memory for Stateful Tool-Using LLM Agents

TL;DR

ClawVM introduces a virtual memory layer for state management in tool-using LLM agents, ensuring deterministic residency and durability with minimal overhead.

Contribution

It presents a novel virtual memory system that enforces state invariants and deterministic management in LLM agent harnesses.

Findings

Eliminates policy-controllable faults when within token budget

Adds median <50 microseconds overhead per turn

Validated across synthetic, real, and stress test workloads

Abstract

Stateful tool-using LLM agents treat the context window as working memory, yet today's agent harnesses manage residency and durability as best-effort, causing recurring failures: lost state after compaction, bypassed flushes on reset, and destructive writeback. We present \textsc{ClawVM}, a virtual memory layer that manages state as typed pages with minimum-fidelity invariants, multi-resolution representations under a token budget, and validated writeback at every lifecycle boundary. Because the harness already assembles prompts, mediates tools, and observes lifecycle events, it is the natural enforcement point; placing the contract there makes residency and durability deterministic and auditable. Across synthetic workloads, 12 real-session traces, and adversarial stress tests, \textsc{ClawVM} eliminates all policy-controllable faults whenever the minimum-fidelity set fits within the token budget, confirmed by an offline oracle, and adds median <50 microseconds of policy-engine overhead per turn.