Real SCE-UA by default
aiswmm calibrate runs the spotpy engine with live progress, unit-magnitude guards, convergence data, best parameters, trials, and candidate audit artifacts.
Runtime & workflow · v0.7.6
From modelling intent to a self-describing run.
Natural-language coordination sits above deterministic EPA SWMM execution. Goals, environments, tools, skills, artifacts, QA, calibration, and audit stay visible at every stage.
Why this project exists
Stormwater modelling is rarely one command. A typical SWMM project can involve GIS preprocessing, rainfall formatting, parameter assignment, network assembly, INP construction, model execution, QA checks, plots, calibration, uncertainty analysis, and reporting.
Agentic SWMM provides a middle path: natural-language orchestration with deterministic SWMM execution, explicit provenance, project memory, and verification-first modelling.
The goal is not to replace SWMM or the modeller, but to make SWMM-based modelling easier to reproduce, audit, remember, and trust.
What changed in v0.7.6
A runtime that can explain the run itself.
The July 2026 release adds operational capabilities without weakening the evidence contract.
Release notes →Goal, agent, and environment captured
Each session stores the verbatim goal, lifecycle, provider and model, prompt and tool-schema hashes, per-skill hashes, permissions, platform, version, and git commit.
Canonical numbered stages
New runs consistently use 05_builder, 06_runner, 07_qa, 08_plot, 09_audit, 10_upstream, and 11_review.
What makes it different
Six surfaces, one evidence contract
Fast, explicit onboarding
Use the one-line installers, PyPI package, or a pinned Docker image. For agent runtimes, install the portable skills separately with one command.
Agent-guided, solver-grounded
Agents plan and route work while EPA SWMM execution remains deterministic, inspectable, artifact-based, and runnable from the CLI.
19 portable skills
GIS, climate, building, network, running, plotting, design review, real calibration, uncertainty, water quality, audit, memory, and orchestration stay modular.
56 typed tools
MCP interfaces expose typed modelling actions while planners are constrained by routing rules, input requirements, QA gates, and stop conditions.
Verification-first provenance
Build, run, QA, plot, audit, upstream, and review stages emit traceable artifacts before outputs are presented as evidence.
Supervised memory and skill evolution
Raw audit notes remain the source record; curated memory can surface patterns and propose skill changes, but acceptance still requires human review and benchmark verification.
The workflow
Three connected layers
The workflow has three connected layers: execution, modeling memory, and controlled skill evolution. Natural-language requests can trigger reproducible SWMM actions; audited artifacts update human-readable and machine-readable memory; repeated patterns can produce skill-refinement proposals that still require human review and benchmark verification.
What a run can produce
- generated or supplied SWMM input files such as
model.inp - SWMM report and binary outputs such as
.rptand.out session.yaml, agent snapshots, environment fingerprints, manifests, command traces, QA summaries, and parsed peak-flow metrics- rainfall-runoff figures, real SCE-UA convergence and calibration summaries, best parameters, trials, and fuzzy uncertainty summaries
- audit records:
experiment_provenance.json,comparison.json, andexperiment_note.md - Obsidian-ready modelling notes plus raw and curated modelling-memory summaries