This is a Preprint and has not been peer reviewed. This is version 1 of this Preprint.
Agentic Modelling Pipeline: Reproducible Rapid Stormwater Modelling Management System with OpenClaw
Downloads
Supplementary Files
Authors
Abstract
Configuring urban hydrological models, such as SWMM, for operational or real-time modelling remains onerous for many models. We propose an Agentic SWMM workflow, which embeds ‘Skills’ and model context protocols to automate model configuration, execution, and extract and plot quantities of interest. To ensure that the entire Agentic SWMM workflow is auditable and reproducible, each run will create an independent manifest file that documents the SWMM version used, parameter mappings, input file hashes, and quality gate (e.g., continuity diagnostics). We demonstrate the workflow on the Tod Creek watershed (located on the Saanich Peninsula, British Columbia) and verified the equivalence between EPA SWMM® command line execution and model context protocol triggered SWMM execution, and between SWMM GUI workflow and the Agentic SWMM workflow. Validation was demonstrated through identical peak. The overall automation workflow (Skills and model context protocol setting), along with plotting services and evaluation protocols are available as Open Source to support a reproducible Agentic SWMM workflow triggered by natural language.
DOI
https://doi.org/10.31223/X5F47G
Subjects
Civil and Environmental Engineering, Civil Engineering, Computational Engineering, Engineering, Environmental Engineering, Hydraulic Engineering
Keywords
Hydrological Modelling, Agentic AI, SWMM, MCP
Dates
Published: 2026-03-18 03:04
Last Updated: 2026-03-18 03:04
License
CC BY Attribution 4.0 International
Additional Metadata
Conflict of interest statement:
None
Data Availability:
https://github.com/Zhonghao1995/agentic-swmm-workflow
Metrics
Views: 14
Downloads: 0
There are no comments or no comments have been made public for this article.