GraphBit: A Graph-based Agentic Framework for Non-Linear Agent Orchestration

GraphBit：一种用于非线性智能体编排的基于图的智能体框架

Abstract: Agentic LLM frameworks that rely on prompted orchestration, where the model itself determines workflow transitions, often suffer from hallucinated routing, infinite loops, and non-reproducible execution. We introduce GraphBit, an engine-orchestrated framework that defines workflows explicitly and deterministically as a directed acyclic graph (DAG).

摘要： 依赖提示词编排（即由模型自身决定工作流转换）的智能体大模型框架，往往会遭遇路由幻觉、无限循环以及执行不可复现等问题。我们推出了 GraphBit，这是一个由引擎驱动的框架，它将工作流显式且确定性地定义为有向无环图（DAG）。

Unlike prompted orchestration, agents in GraphBit operate as typed functions, while a Rust-based engine governs routing, state transitions, and tool invocation, ensuring reproducibility and auditability. The engine supports parallel branch execution, conditional control flow over structured state predicates, and configurable error recovery.

与提示词编排不同，GraphBit 中的智能体以类型化函数的形式运行，而基于 Rust 的引擎则负责管理路由、状态转换和工具调用，从而确保了执行的可复现性和可审计性。该引擎支持并行分支执行、基于结构化状态谓词的条件控制流以及可配置的错误恢复机制。

A three-tier memory architecture consisting of ephemeral scratch space, structured state, and external connectors isolates context across stages, preventing cascading context bloat that degrades reasoning in long-running pipelines.

其三层内存架构由临时暂存空间、结构化状态和外部连接器组成，能够隔离各阶段的上下文，防止在长流程管道中因上下文膨胀而导致的推理能力下降。

Across GAIA benchmark tasks spanning zero-tool, document-augmented, and web-enabled workflows, GraphBit outperforms six existing frameworks, achieving the highest accuracy (67.6 percent), zero framework-induced hallucinations, the lowest latency (11.9 ms overhead), and the highest throughput. Ablation studies demonstrate that each memory tier contributes measurably to performance, with deterministic execution providing the greatest gains on tool-intensive tasks representative of real-world deployments.

在涵盖零工具、文档增强和网络启用工作流的 GAIA 基准测试任务中，GraphBit 的表现优于六种现有框架，实现了最高的准确率（67.6%）、零框架诱导幻觉、最低的延迟（11.9 毫秒开销）以及最高的吞吐量。消融研究表明，每一层内存架构都对性能有显著贡献，其中确定性执行在代表实际部署的工具密集型任务中带来了最大的性能提升。