DETECT v2

A NiceGUI webapp and command line evaluator over the DETECT v2 SysML v2 models in model/, using Syside Automator.

The models are the source of truth. Use cases, questions, answers, help text, lifecycle phases, job series, criteria, requirements, their ordering and tool mappings are all read out of the models, so extending DETECT means editing the models rather than the Python.

Running it

Syside needs a Pro licence as soon as it is imported, so each uv run command below needs one. Provide it through one of these environment variables:

• SYSIDE_LICENSE_KEY, holding the key itself
• SYSIDE_LICENSE_FILE, holding the path to a .lic certificate

Keep the licence out of this repository; .gitignore excludes *.lic.

Install uv, then:

uv sync
uv run python webapp_main.py      # webapp at http://localhost:8080, or on $PORT
uv run python detect.py           # command line, writes CSVs to Output/
uv run python -m pytest tests/    # tests

Running it this way needs no Docker.

The evaluator loads every .sysml file in model/.

The command line path reads your answers out of the model files themselves: set the sizing attributes in model/sizing_input.sysml away from TBD, and set the boolean flags in model/lifecycle_phase_input.sysml and model/job_series_input.sysml. The webapp writes the same answers into an in-memory copy of the model instead, so it leaves the files alone.

The webapp's "Feedback" links send email to the address and attention line declared in model/feedback.sysml. Edit that file to change where feedback goes.

Optionally, in a dev container

.devcontainer/ describes a VS Code dev container with Python, uv, the dependencies and the Syside extension installed. Using it is optional, and it needs two things the steps above do not:

1. Docker. On Windows or macOS, install Docker Desktop; on Windows it also needs WSL 2, which its installer sets up. On Linux, install Docker Engine. Docker Desktop requires a paid subscription for use in larger organisations and in government; see its licence terms.
2. The VS Code Dev Containers extension.

Set SYSIDE_LICENSE_KEY on your machine, start VS Code from an environment where it is set, then run Dev Containers: Reopen in Container. The container reads the key from that environment.

What it produces

File	Contents
requirements.csv	id, value, description — ecosystem requirements applicable at the calculated size
criteria.csv	id, value, criteria, context — tool evaluation criteria applicable at the calculated size
uc2_tools.csv	tool_type, required_by — tool types required by the selected lifecycle phases
uc3_tools.csv	tool_type, required_by — tool types required by the selected job series

The tool type files are named after the use case's short name in the model, so a fourth use case gets its own file without a change to the Python. The value column is the weight the model assigns, rounded to four decimal places on output only.

Layout

• model/ — the SysML v2 models, which are the source of truth.
• detect.py — the evaluator, and the command line entry point. Nothing here reimplements a filtering rule, a weight or a tool mapping; all of it is evaluated from the models.
• webapp_main.py — the webapp: a landing page, a configuration page with one tab per use case, tables and CSV downloads.
• README_web.md — the text of the landing page.
• tests/ — tests over the evaluator, run against the models in model/.

Container image

This section is only for deploying the webapp; running it locally does not need it.

docker build -t detect-webapp .

Dependencies are managed by uv: pyproject.toml declares them and uv.lock pins every package by version and hash. The image installs with uv sync --frozen, which fails rather than silently relocking.

The image contains no licence. SYSIDE_LICENSE_FILE points at /secrets/syside-license.lic, which is expected to be mounted at runtime, for example:

docker run -p 8080:8080 -v /path/to/syside-license.lic:/secrets/syside-license.lic:ro detect-webapp

Do not pass the licence through an ENV instruction or a build argument: those are recorded in the image and can be read with docker history. .dockerignore excludes *.lic from the build context for the same reason.