vibeflow-scilab

name	vibeflow-scilab
description	Use when preparing, running, debugging, or extending VibeFlow/VibeCAE Scilab numerical computing workflows from 0-caseDict/caseDict, including scripts, functions, optimization, control, signal processing, data reduction, plotting, Xcos handoff, and reusable scientific-computing utilities.

VibeFlow Scilab

This skill is responsible for using Scilab as the numerical computing, control, optimization, data processing, and engineering scripting backend for VibeCAE.

• Usage mode: generate and run Scilab scripts from caseDict to complete parameter calculations, reduced-order models, control analysis, optimization, data processing, or chart output.
• Development mode: develop reusable .sci function libraries, batch scripts, Xcos/control-tool interfaces, or Studio invocation bridges.

When To Use

• Matrix calculations, control systems, optimization, signal processing, statistical processing, or engineering calculation scripts are needed.
• Lightweight numerical analysis of CFD/FEM/MD results is needed, but the full Python ecosystem is not required.
• The user already has Scilab scripts or wants to connect MATLAB-style workflows to VibeCAE.
• .sce batch scripts, .sci functions, charts, and tables need to be generated.

First Decision: Usage Mode Or Development Mode

• One-off calculations, parameter tables, and post-processing plots for the current case: use usage mode.
• General function libraries, templates, Studio adapters, Xcos models, or batch frameworks: use development mode.
• If Scilab commands are unavailable or the graphics backend is abnormal: switch to vibeflow-toolchain first.
• If Python is the more suitable backend, prefer the existing Python post-processing workflow.

What To Read First

• Parameters, objective functions, inputs, outputs, and post-processing requirements in 0-caseDict/caseDict.
• references/scilab-playbook.md
• references/scilab-practical-tips.md: software knowledge base, key concepts, failure interpretation, and adapter design points.
• Existing .sce, .sci, and data tables under C-run/scilab/, D-post/, and Ref/.

Inputs

• 0-caseDict/caseDict
• CSV, TSV, DAT, MAT, result logs, and monitor files
• .sce scripts and .sci functions

Outputs

• C-run/scilab/case-*/main.sce
• C-run/scilab/case-*/functions/*.sci
• C-run/scilab/case-*/logs/
• D-post/scilab_* charts, tables, and intermediate data
• Post-processing records in caseDict when needed

Recommended Structure

C-run/scilab/
└── case-1/
    ├── main.sce
    ├── functions/
    ├── data/
    ├── results/
    ├── logs/
    ├── Allrun
    └── Allclean

Usage-Mode Workflow

1. Determine input data, output metrics, units, and calculation objectives from caseDict.
2. Write main.sce, anchoring paths to the current case instead of relying on the interactive current directory.
3. Put reusable functions under functions/ and load them explicitly at the beginning of the script.
4. Run Scilab in batch mode and save logs.
5. Output CSV/PNG/SVG/PDF results and check that the files exist.
6. Hand charts and calculation results to vibeflow-postprocess or the report skill.

Development-Mode Workflow

1. Clarify the development target: function library, optimization template, control-analysis script, Xcos export, or Studio adapter.
2. Write the development note first, recording function interfaces, inputs, outputs, defaults, error handling, and a minimal example.
3. Keep scripts runnable in a headless environment.
4. Add minimal numerical validation for key functions, such as known analytical solutions, benchmark data, or dimension checks.
5. If coupled with a solver, clearly define data-file formats and refresh timing.

Core Rules

• Scilab scripts must not hide key physical assumptions; units and data sources must be clear.
• Use .sci for reusable functions and .sce for one-off workflows.
• Graphics output must have deterministic paths and must not depend on manual saving from GUI windows.
• For large files, distinguish memory requirements and preprocess or read in chunks when needed.

Validation

• The Scilab CLI can run.
• main.sce executes successfully in batch mode and the log has no errors.
• Output files exist and are non-empty.
• Key numerical results pass a simple sanity check.

Fallback Rules

• Numerical model or objective function unclear: return to vibeflow-case-dict.
• Environment unavailable: return to vibeflow-toolchain.
• Results are intended for a report but evidence is insufficient: switch to vibeflow-postprocess or vibeflow-report.