Annotate Control Panel

Annotate Control Panel

This skill writes the Wolfram control-panel (Explore) annotation into a Modelica .mo file so a model carries its own interactive panels — the sliders, checkboxes, input fields and popup menus that Simulation Center's Explore view shows for driving parameters and start values, plus the plots to display:

annotation(__Wolfram(ControlPanels(
  Panel(
    identifier = "id1",
    title = "Controller tuning",
    elements = {
      Slider(variable = Kp, label = "Proportional gain", min = 0, max = 100),
      Slider(variable = Kd, min = 0, max = 20, scale = Log()),
      Checkbox(variable = useFeedforward, label = "Feedforward")},
    figures = {"response"})));

It is a self-contained source transform: Claude assembles a control-panel spec (JSON) — from the user's request, or suggested from the model's parameters — and the Python engine renders it to the __Wolfram(ControlPanels(...)) vendor annotation and splices it into the class annotation. The engine is pure Python (standard library only); WSMKernelX is used only afterwards as a validation gate.

This is a vendor-specific annotation (Modelica spec §18.1): other tools ignore it and preserve it on save, and it does not affect flattening — it just tells System Modeler how to build the Explore panels. It sits directly inside annotation(...), as a sibling of experiment and Documentation (never nested inside Documentation).

It is idempotent: a class that already has control panels is skipped unless you pass --force (which strips the old ControlPanels block and regenerates). By default it prints a dry-run diff; nothing is written until --write.

> --force is destructive — ask the user first. It deletes the class's existing control panels > before regenerating, so any hand-tuned panels there are lost. Don't pass --force on a class that > already carries control panels without confirming that discarding them is intended. Review the > dry-run diff (no --write) to see exactly what would be removed.

The control-panel spec (JSON)

A model has a list of panels; each panel has a list of controls and an optional list of figures (plot identifiers to show). Only fields you set are emitted:

{
  "panels": [
    {
      "identifier": "id1",
      "title": "Controller tuning",
      "controls": [
        {"type": "slider",     "variable": "Kp", "label": "Proportional gain", "min": 0, "max": 100},
        {"type": "slider",     "variable": "Kd", "min": 0, "max": 20, "scale": "Log", "showInputField": false},
        {"type": "checkbox",   "variable": "useFeedforward", "label": "Feedforward"},
        {"type": "inputField", "variable": "mass"},
        {"type": "popupMenu",  "variable": "mode", "items": [{"value": "1", "label": "Fast"}, {"value": "2"}]}
      ],
      "figures": ["response"]
    }
  ]
}

Control type is chosen from the variable's kind:

• slider — a Real/Integer parameter or start value. Requires numeric min and max.

Optional "scale": "Log" for a logarithmic slider, and "showInputField": false to hide the numeric field beside the handle (it defaults to shown).

• inputField — a typed value for a Real/Integer/String.
• checkbox — a Boolean.
• popupMenu — an enumeration or Boolean. Give items as {"value": , "label": "…"};

value is emitted verbatim, so use a bare token for a number/enumeration ("2", "MyEnum.fast") and include quotes for a String value ("\"idle\"").

Rules the engine enforces:

• Each control's variable is a component reference (e.g. Kp or controller.gain[1]) — the

parameter/start value to control. It is emitted unquoted, as System Modeler expects.

• A slider must have finite numeric min/max.
• Each panel needs a unique, non-empty identifier and a title.
• figures entries are Figure identifiers that must exist in the model's stored figures

(written by the annotate-modelica-plots skill); the first listed is the panel's default plot.

Running the annotator

The annotator lives in the ControlPanel/ package next to this file. Run it as a module from this skill's directory so the package is importable:

cd ""
python3 -m ControlPanel.main --file "" --analyze

Use python instead of python3 on Windows if that's what's on PATH. Only --file (and --spec, --sim-file) must be absolute or correctly-relative paths.

Cross-platform launcher (for the validate step)

> The validate step below calls the shared scripts/wsm_run.py. Read > [the shared-conventions appendix at the end of this file](#appendix-shared-conventions-for-the-modelica-skills) (launcher resolution, shell rules, temp dir, > JSON-array output, MSL 4.x notes) before it. In a normal install the launcher is > ../scripts/wsm_run.py (` is the shared scripts/` folder); if that path doesn't > exist, see [Appendix → Locating the launcher](#locating-the-launcher).

Workflow

1. Identify the model file and target class

The user gives a .mo path. Pick a concrete instantiable model (Package.Model, not the package) — see [Appendix → Picking the model name](#picking-the-model-name). Control panels belong on the model you actually simulate.

2. Analyze — see classes, panel status, and controllable parameters

python3 -m ControlPanel.main --file "" --analyze

Lists every class, flagging HAS control panels, and lists the target class's controllable parameters with their kind (numeric / boolean / string / other) and default. Present this and confirm scope. Target a specific class with --class .

3. Choose the controls

Expose the parameters a user would actually want to turn — gains, setpoints, masses, switches, mode selectors — not every constant in the model. Prefer a small, meaningful panel.

Mode A — user-specified. The user names the parameters (and ranges). Build the spec JSON directly from their request and skip to step 4.

Mode B — suggest from the model's parameters. When the user doesn't name them, let the tool propose a starting spec from the class's parameter declarations:

python3 -m ControlPanel.main --file "" --class "" --suggest --max-controls 12 > "/panels.json"

It picks a control per parameter by type (slider for numeric with a first-guess range around the default, checkbox for Boolean, input field otherwise) — a starting point. Refine panels.json: drop parameters that shouldn't be exposed, set real slider min/max, add labels, switch an enumeration to a popupMenu with items, and split into multiple panels if useful.

Optional — check against a built model. If the model has been built or simulated, pass the .sim init file (beside the result .mat, same stem) so the tool confirms each controlled variable exists and is runtime-tunable, warning on structural (constant-folded) parameters that can't be changed from a panel without a rebuild:

python3 -m ControlPanel.main --file "" --class "" \
  --spec "/panels.json" --annotate --sim-file "/_wsm_simulate_temp/_res.sim"

3b. Labels, ranges, and linking plots

• Label — give each control a short label (the parameter's description is a good default). It

is what the user sees next to the slider/box.

• Range — set slider min/max to the span worth exploring, not just around the default. Use

"scale": "Log" when the interesting range spans orders of magnitude.

• Title — name the panel for what it controls (e.g. "Controller tuning"), not "Panel 1".
• Plots — list figures (by Figure identifier) so the panel shows the relevant plot as you

drag. The identifiers must already exist in the model's stored figures; add them first with the annotate-modelica-plots skill if they don't. The first listed figure is the default plot.

4. Preview — dry-run diff

python3 -m ControlPanel.main --file "" --class "" --spec "/panels.json" --annotate

Shows the unified diff without writing. Warnings surface unknown/structural variables, control-type mismatches, and figure ids with no stored Figure. --force regenerates over existing panels.

5. Apply — write in place

python3 -m ControlPanel.main --file "" --class "" --spec "/panels.json" --annotate --write

Re-running without --force is a no-op for classes that already have control panels (idempotent).

6. Validate — confirm the model still flattens (validate-modelica gate)

The annotation must not change the flatten result.

python3 "/wsm_run.py" --mode validate --model "" --name "" --timeout 90

Parse _wsm_validate_temp/validate.out.json (a JSON array — take [0]; field reference: [Appendix → Reading the JSON output](#reading-the-json-output)) and check status.flatten == "Pass". A vendor annotation is ignored by flattening, so a failure here that wasn't present before means the annotation is malformed — report it. Then remove the temp dir.

Notes and edge cases

• Pick an instantiable model for the kernel step, not the package — use a nested model's full

dotted name. See [Appendix → Picking the model name](#picking-the-model-name).

• Control panels vs. plots. This skill writes the interactive __Wolfram(ControlPanels(...))

panels; the annotate-modelica-plots skill writes the stored Documentation(figures={...}) plots. A panel's figures list just references figure identifiers from that annotation — add the plots there first, then link them here.

• Bind to parameters and start values. Controls change the start value of a variable or a

parameter value — the things you set before (or, in real-time mode, during) a run. A control on a purely computed variable can't do anything useful.

• Structural parameters. A parameter that is constant-folded (used in array sizes, conditional

components, etc.) can't be changed from a panel without rebuilding. Pass --sim-file to get a warning when a controlled variable is structural or absent from the built model.

• Enumerations and popup values. popupMenu items values are emitted verbatim; use the bare

enumeration/number token ("value": "2"), and quote a String value ("value": "\"idle\"").

• Idempotency / regen. Default runs never duplicate a control-panel block; use --force to

regenerate after editing the spec — but it discards the existing panels, so confirm with the user first (see the caution near the top). --force preserves other content in annotation(...) (e.g. experiment, Documentation).

• MSL dialect (this toolchain ships MSL 4.x). The annotation doesn't affect flattening, but if

the model uses 3.2 names it fails the gate for unrelated reasons — see [Appendix → MSL 4.x dialect](#msl-4x-dialect).

---

Appendix: shared conventions for the Modelica skills

> Shared by every Modelica skill that drives WSMKernelX through the > bundled launcher; inlined here at release time. For the CLI/option > reference, environment variables (WSM_HOME, WSM_VSDEVCMD), > install discovery, and the analysis scripts, see > [../scripts/README.md](../scripts/README.md).

Locating the launcher

` (used throughout the skills) is the shared scripts/` folder. Some installs symlink the skill directories without it, so resolve it in this order and use the first that exists:

1. $WSM_SKILLS_SCRIPTS (bash) or $env:WSM_SKILLS_SCRIPTS (PowerShell), if set.
2. ../scripts relative to the skill directory — in a normal install

../scripts/wsm_run.py already exists, so use that path directly; do not run a shell probe to "resolve" it.

1. The repo checkout you installed from, e.g. .../agentskills/scripts.
2. Last resort, search the home directory:

• PowerShell: Get-ChildItem $HOME -Recurse -Filter wsm_run.py -ErrorAction SilentlyContinue | Select-Object -First 1
• bash/zsh: find ~ -name wsm_run.py -path '*scripts*' 2>/dev/null | head -1

If only #4 finds it, the install is missing the scripts/ link — tell the user to run install.sh (or install.ps1) from the repo, which links scripts/ too.

Shell and Python

On Windows, use PowerShell. The Git-Bash/cygwin layer may be broken (even ls/find can be absent, giving a misleading "exit 127 / command not found"). Run wsm_run.py with python (not python3); those calls are single-line and shell-agnostic. For cleanup use Remove-Item -Recurse -Force, not rm -rf. On macOS/Linux any POSIX shell is fine and python3 is the usual name.

Let the launcher own .mos/.bat and paths

Do not hand-write .mos scripts, .bat files, or hardcode install/compiler paths. The bundled scripts/wsm_run.py handles every OS difference — it finds the System Modeler install and kernel binary (macOS / Windows / Linux), finds and loads the right MSL files, generates the .mos, and runs the kernel with a working compiler environment per platform (system clang/gcc on macOS/Linux; the Visual Studio dev environment via VsDevCmd.bat on Windows). See [../scripts/README.md](../scripts/README.md) for WSM_HOME, the Windows compiler prerequisites, and the full option table.

When the install or compiler isn't found

The launcher searches each OS's standard install locations. If it prints ERROR: Could not locate a Wolfram System Modeler installation, the install is in a non-standard place — ask the user for it and re-run with --wsm-home "" (or have them set WSM_HOME).

Building and simulating also need a C++ toolchain:

• Windows: Visual Studio Build Tools. The launcher locates VsDevCmd.bat

itself; if it reports the compiler environment is missing, pass --vsdevcmd "" (or set WSM_VSDEVCMD) and make sure Build Tools are installed.

• macOS: the Xcode command-line tools (xcode-select --install).
• Linux: gcc/g++.

Run python3 "/wsm_run.py" --mode info to see what the launcher discovered.

Temporary directories

The launcher works in a _wsm__temp/ directory next to the .mo file (_wsm_validate_temp/, _wsm_simulate_temp/, _wsm_diagnose_temp/) and leaves its outputs there. Tell the user, e.g.: "Working in temporary directory _wsm__temp/. This will be deleted afterwards." Pass --tempdir to reuse one directory across models in a session.

Clean up by removing the whole directory — use the user's shell:

rm -rf "/_wsm__temp"          # macOS / Linux
# PowerShell: Remove-Item -Recurse -Force "\_wsm__temp"

Picking the model name

• The user may provide a path to a .mo file, or you may already be working with

one in context.

• Extract the model name: the identifier after model on the first non-comment

line, e.g. model FooBar → FooBar. The filename does not always match the model name — parse the actual model/package declaration.

• For packages or nested models, use the top-level model name.
• Pick an instantiable model, not a package, for any kernel call. A package

cannot be validated or simulated ("Invalid instantiation … is a package") — use a nested model's full dotted name, e.g. Package.Model.

• Pass an absolute path to --model (relative paths break as the working

directory shifts between calls).

Directory-form (multi-file) libraries

A directory-form library stores one class per file with a package.mo at each level. You cannot validate such a class by handing the launcher only its own .mo file — the class's within Lib; clause needs the whole package loaded, and loading the single file alone fails with Internal error: ... expandLibNode: Unknown library: Lib. Instead point --model at the library folder (or its top package.mo, or any class file inside it) and pass the full dotted class name via --name:

python3 "/wsm_run.py" --mode validate \
  --model "/abs/path/InvertedPendulum" \
  --name InvertedPendulum.Controller

The launcher resolves any of

…

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: WolframResearch
• Source: WolframResearch/system-modeler-ai-toolkit
• License: MIT
• Homepage: https://www.wolfram.com/system-modeler/

Install and usage instructions live in the source repository linked above.