KAIJU: An Executive Kernel for Intent-Gated Execution of LLM Agents
For developers of LLM-based autonomous agents, KAIJU offers a system-level abstraction that improves security and execution efficiency over the ReAct paradigm.
KAIJU introduces an executive kernel that decouples LLM agent planning from execution, enabling parallel tool dispatch and intent-gated security. It achieves latency parity with ReAct on moderate tasks and outperforms on parallelizable queries, while providing behavioral guarantees against prompt injection.
Tool-calling autonomous agents based on large language models using ReAct exhibit three limitations: serial latency, quadratic context growth, and vulnerability to prompt injection and hallucination. Recent work moves towards separating planning from execution but in each case the model remains coupled to the execution mechanics. We introduce a system-level abstraction for LLM agents which decouples the execution of agent workflows from the LLM reasoning layer. We define two first-class abstractions: (1) Intent-Gated Execution (IGX), a security paradigm that enforces intent at execution, and (2) an Executive Kernel that manages scheduling, tool dispatch, dependency resolution, failures and security. In KAIJU, the LLM plans upfront, optimistically scheduling tools in parallel with dependency-aware parameter injection. Tools are authorised via IGX based on four independent variables: scope, intent, impact, and clearance (external approval). KAIJU supports three adaptive execution modes (Reflect, nReflect, and Orchestrator), providing progressively finer-grained execution control apt for complex investigation and deep analysis or research. Empirical evaluation against a ReAct baseline shows that KAIJU has a latency penalty on simple queries due to planning overhead, convergence at moderate complexity, and a structural advantage on computational queries requiring parallel data gathering. Beyond latency, the separation enforces behavioural guarantees that ReAct cannot match through prompting alone. Code available at https://github.com/compdeep/kaiju