sysml2kit: a Python SDK for SysML v2
- Python
- pydantic
- typer
- MCP
- SysML v2
Problem
SysML v2 finally gives systems engineering a textual notation and a standard API, but the Python tooling around it stops at parsing and transport. There was no layer where a requirement with a threshold, the architecture that satisfies it, and the analysis that verifies it live in one model that software and AI agents can query, check, and execute.
Approach
sysml2kit (Apache-2.0, on PyPI) models a documented subset of SysML v2 as pydantic objects and
emits standard textual notation plus Systems Modeling API JSON, preserving out-of-profile elements
opaquely. On top of that sit traceability queries, ten validation rules, element-level diffs, and a
verification runner: analyses bind to engines resolved by name from an entry-point registry, and
sibling bindings labeled with fidelity and cost form a ladder whose escalate policy spends a
wall-clock budget on the requirements with the thinnest margins. An MCP server exposes nine tools
so agents query the model instead of parsing files.
Result
One command takes a SysML v2 model to executed verification with margins and provenance, at 0.4.0 still pre-alpha and moving. Domain content stays outside the kit: an RF model library and engine integrations with phased-array-systems and AEDL build on it. I wrote up the binding convention and the 0.4.0 fidelity-budget release on the blog.