agentic-mbse

Domain-agnostic MBSE toolkit for AI-assisted systems engineering.

Installation

pip install -e .

Usage

Validate SysML Models

agentic-mbse validate models/

Initialize Project

agentic-mbse init

This creates:

• .claude/commands/ - MBSE workflow commands (/design-model, /implement-model, etc.)
• .claude/agents/ - Specialized AI agents for SysML and documentation
• .claude/skills/ - Skills including /record-learning for capturing insights
• modeling_pm/ - Project management structure (backlog, active work, learnings)
• SOURCE_INDEX.md - Domain knowledge source configuration

Learning Feedback Loop

During modeling sessions, capture insights and discoveries with /record-learning:

/record-learning

This triggers reflection on the current conversation, identifying:

• Import patterns discovered
• Syntax gotchas resolved
• Error interpretations learned
• Workarounds implemented
• Best practices identified

Learnings are stored in modeling_pm/learnings/RAW_LEARNINGS.md for later review and formalization into documentation.

Development

pip install -e ".[dev]"
pytest tests/

Maintaining Documentation Corpus

The docs/sysmlv2/ directory contains indexed specifications and standard library files that power the documentation agents. When syside or SysML v2 specs are updated, regenerate using these scripts:

Script	Purpose	When to Run
scripts/sync_stdlib.py	Sync standard library from syside package	After syside upgrade
scripts/generate_index.py	Generate INDEX.md for spec documents	After extracting new PDFs

# After upgrading syside
python scripts/sync_stdlib.py --force

# After extracting new spec PDFs to docs/sysmlv2/
python scripts/generate_index.py docs/sysmlv2/SysML_KerMLSpec/
python scripts/generate_index.py docs/sysmlv2/SysML_Spec_v2_Part1/
# ... etc for each spec directory

See .modeling_pm/backlog/epic_documentation-discoverability.md for full documentation of the indexing approach.