graph-based tool orchestration

This work addresses the lack of a unified framework in current LLM agent workflows, which hinders method comparison and reproducibility. To resolve this, we propose the Agent Computation Graph (ACG) framework, which models workflows as computation graphs and adopts “structure determines timing” as a core principle. The framework explicitly distinguishes between reusable templates, runtime instance graphs, and execution traces, enabling a systematic categorization of static and dynamic optimization approaches. Through a comprehensive literature review and conceptual modeling, we develop a multidimensional evaluation framework that integrates structural properties, establishes precise terminology, and defines standardized evaluation criteria. This foundation supports a reproducible and highly comparable research paradigm for optimizing LLM agent workflows.

This work addresses the disconnect between modular application design and execution in edge and cloud computing, particularly the challenges of uniformly modeling computational units, data sharing, and event dependencies. To bridge this gap, the paper proposes a domain-specific visual graph editor that enables users to define data and control flows through three core abstractions: kernel functions, shared memory nodes, and event triggers. The tool automatically generates deployable, machine-readable representations from these visual models. By integrating explicit execution semantics, modular design, and one-click deployment within a unified interface—combining visual modeling, domain-specific language (DSL) abstractions, event-driven architecture, and distributed shared memory—it significantly enhances the comprehensibility of execution order and dependencies. Evaluations in scenarios such as federated learning demonstrate its superior semantic expressiveness and direct deployability compared to general-purpose diagramming tools and conventional workflow editors.

Existing intelligent agents often suffer from high system fragility and substantial execution overhead in multi-tool coordination due to inadequate scheduling mechanisms. This work proposes a hierarchical orchestration paradigm that obviates the need for fine-grained dependency graphs by providing coarse-grained global guidance, coupled with context-constrained intra-layer execution. A pattern-aware local reflection mechanism is introduced to enable runtime error detection and repair without triggering costly global replanning. The approach significantly enhances the robustness of tool invocation while reducing execution complexity and resource consumption, yielding a lightweight and reusable tool orchestration component.

This study addresses the challenge of automating workflows in complex industries—such as logistics, healthcare, and construction—where processes are fragmented across heterogeneous tools and involve multi-party collaboration. The work proposes orchestration as a core abstraction to enable effective automation by dynamically coordinating multi-step tasks, enforcing domain-specific constraints, managing human approvals, and integrating legacy systems. It introduces the novel concept of “orchestration bottlenecks” and develops a theoretical framework that unifies multi-agent systems, workflow modeling, constraint reasoning, and human–AI collaboration, while exposing critical gaps in current multi-agent approaches at the orchestration level. Based on distinct sources of operational friction across domains, the paper advocates for targeted architectural safeguards—such as constraint enforcement or explainability—and phased implementation strategies to provide actionable pathways for automation in complex operational environments.

This study addresses the challenges of maintaining consistency across heterogeneous schema languages—such as JSON Schema, XSD, and SHACL—during multilingual data model evolution, where fragmented converters, variable quality, and information loss impede reliable interoperability. The work proposes a novel approach that models schema languages and black-box converters as nodes and directed edges in a graph, enabling composable and evaluable conversion path orchestration. By integrating graph-based search, quality-aware ranking (combining agent-assisted and human evaluation), and failure backtracking, the method supports automated, reproducible cross-language schema transformation. The resulting open-source toolchain, Schema Conversion Orchestrator, integrated into the MetaConfigurator platform, successfully produced valid outputs for 43 out of 60 real-world tasks and precisely identified missing ecosystem components in the remaining 17, thereby delineating the current boundaries of schema conversion capabilities.

This work addresses the challenges of constructing and managing generative AI agent systems for long-horizon, stateful, multi-step business processes by proposing a graph-structured workflow design methodology. Leveraging the LangGraph framework, it explicitly models core mechanisms such as state management, conditional routing, and human-in-the-loop interventions. The approach is instantiated in three representative applications: SQL analysis with repair loops, retrieval-augmented generation gated by evidential validation, and human-AI collaborative policy review supporting interruption and checkpoint-based recovery. By treating behaviors like routing, pausing, and audit trails as explicit product features rather than implicit prompt logic, this study not only delineates the applicability boundaries of LangGraph in high-complexity workflows but also substantially enhances system controllability, reliability, and auditability in real-world operational settings, establishing a reusable engineering paradigm.

This work addresses the challenge that large language models struggle to efficiently plan tool usage in complex tasks due to implicit reasoning and dynamic environmental changes. It introduces a novel approach that models tool relationships at the schema level by constructing a tool–schema hypergraph, where each tool is represented as a hyperedge connecting input and output schema nodes. The method further incorporates a task-relevant context graph, a schema-aware task-directed acyclic graph (DAG), and a gap-driven expansion mechanism conditioned on system state to enable precise dynamic planning. Evaluated on the AppWorld benchmark, this framework significantly improves task completion rates while simultaneously reducing redundant API calls, LLM interactions, and token consumption.

Existing approaches rely on tool-level graph representations of historical trajectories, which struggle to generalize to new tool sets and thereby limit the planning capabilities of large language models. To address this, this work proposes a Functional-level Workflow Graph (FWG) that abstracts tool-specific behaviors into functional-level workflows through trajectory uplifting, effectively decoupling workflow planning from tool selection. The framework incorporates a source-gating mechanism and skill-specific rewards, combined with reinforcement learning, to ensure reliable and traceable data flows. Evaluated on two in-distribution and three out-of-distribution benchmarks, the method significantly outperforms current state-of-the-art approaches and demonstrates strong cross-domain generalization to unseen tool sets.