Metadata-Version: 2.4
Name: matrx-assignment
Version: 0.1.0
Summary: Typed planning and durable coordination for resolved variable assignments
Requires-Python: >=3.13
Requires-Dist: pydantic>=2.12
Provides-Extra: dev
Requires-Dist: pytest-asyncio>=0.25.0; extra == 'dev'
Requires-Dist: pytest>=8.3.0; extra == 'dev'
Requires-Dist: ruff>=0.9.0; extra == 'dev'
Description-Content-Type: text/markdown

# matrx-assignment

Typed, payload-agnostic assignment planning and durable coordination.

The library turns a strict plan into immutable work items, then drives those
items through an injected store and executor. Hosts can begin with coordinated
rows (for example, 50 paired `topic` + `research` inputs) and later add policies
without changing the agent or workflow execution systems.

Supported plans:

- `coordinated_rows`: preserve paired values and run every row once.
- `independent_random`: choose every variable independently with unbiased system
  randomness, optionally without repeating a full combination.
- `cartesian`: enumerate combinations, optionally in unbiased random order.

Random plans are materialized once and persisted by the host. Recovery therefore
resumes the same items instead of producing a new draw.
