KSL Application Guides

Step-by-step, user-facing guides for the KSL applications. Most are desktop apps — each guide walks through the app click-by-click with a concrete worked example. Bundle Tools (kslpkg) is a command-line tool, and the KSL Server drives KSL from an AI assistant.

First time? Install the KSL suite — one command installs all these apps, the KSL Server, and kslpkg into a single KSLWork folder, running on your own Java 21 (no build required).

New to these apps? Read Common UI & concepts next — it covers the parts every app shares (models & bundles, the workspace, themes, the run console, reports), so the individual guides don't repeat them.

The apps

Guide	What it's for	Status
Single-Model	Run one model, set its inputs, read a report. The best starting point.	✅ Available
Scenario	Compare several configurations of a model side by side.	✅ Available
Experiment	Vary inputs over a designed (factorial) experiment.	✅ Available
Simopt	Search for the input settings that optimize a response.	✅ Available
Animation	Watch a model run as a visual, replayable animation — capture, run, lay out, replay.	✅ Available
Results	Browse and compare results saved in a simulation database.	✅ Available
Distribution	Fit probability distributions to data.	✅ Available
Bundle Workbench	Package models as bundle JARs in a guided desktop app — open, identify, catalog, validate, assemble.	✅ Available
Bundle Tools	Package models as loadable bundle JARs (kslpkg, command line).	✅ Available

Servers

Prefer to drive KSL from outside a GUI? The server modules expose KSL's capabilities to programs and AI assistants. The KSL Server is the one to use: a single long-running server that gives an AI assistant searchable, tool-driven access to all three surfaces at once — running models, the source code, and the textbook — started from a menu-bar / system-tray app and set up with one click from a web console.

See the KSL Server guide to install it, open the console, connect your assistant with one click, and make a first tool call. It runs headless too, and you can serve just some surfaces (say, textbook search only for a course that isn't modeling yet).

How the apps relate

flowchart TD
    bundle["Model bundle JAR<br/>(Bundle Workbench / kslpkg)"]
    bundle --> single["Single-Model<br/>run one model"]
    bundle --> scenario["Scenario<br/>compare configurations"]
    bundle --> experiment["Experiment<br/>designed experiment"]
    bundle --> simopt["Simopt<br/>optimize inputs"]
    bundle --> animation["Animation<br/>visual replay"]
    single --> db[("Results database<br/>+ reports")]
    scenario --> db
    experiment --> db
    simopt --> db
    db --> results["Results<br/>browse & compare"]
    data["Your data"] --> distribution["Distribution<br/>fit a distribution"]
    distribution -.->|"input models for"| bundle

A model is packaged once as a bundle, then run by the Single, Scenario, Experiment, or Simopt apps — or replayed visually by Animation. Those runs write a results database and reports, which the Results app browses and compares. The Distribution app is the front of the pipeline — it fits distributions to data that feed your models.

For guide authors

The standard structure for every guide is in _TEMPLATE.md.