matadore
Security scanners give you a list. The aim here is a story.
This one is pre-alpha, and the honest version matters more than the pitch. matadore is
published to PyPI and has a full documentation site, but it does not perform scans today. The
entry point, all six input handlers, all five tool plugins and every report method currently
raise NotImplementedError, and the command-line interface is still scaffolding. Treat it as a
design worked out in the open, not a tool to reach for.
What exists is the skeleton: the Pydantic domain models, a dry-run planner that describes exactly what an engagement would touch, a state store over SQLite with optional DuckDB for team-shared runs, a multi-provider LLM client through LiteLLM, and a test suite around all of it.
The intent is adversarial reasoning rather than another checklist — chaining findings into narrative attack paths mapped to MITRE ATT&CK, across domains, network ranges, repositories, GitHub organisations, cloud accounts and container registries, wrapping existing tools rather than replacing them. Two constraints are baked into the models: every finding must carry a verifiable evidence reference, and inference can run entirely locally through Ollama, so nothing has to leave your network.
The diagrams below are the intended architecture and the model layer as designed — the models are real code, the boxes around them mostly are not yet.
Progress is tracked in the open on GitHub.
STATUSPre-alpha — not yet functional
PRICINGFree · MIT
PLATFORMPython package
STACKPython 3.12+, typer, litellm, pydantic
WEBSITEdariomory.github.io/matadore/
GITHUBdariomory/matadore
