noumenal-ai

Engineering + research conventions for the causal-worlds Python package. Load and apply BEFORE writing, refactoring, or reviewing ANY code in this repo (src/, tests/, cli, the package), and before adding a dependency or a design pattern. Encodes Clean Code (Uncle Bob), SOLID-via-Protocols, the earned-patterns rule, the uv/ruff/ mypy/pytest/CI toolchain, src-layout, and the research discipline. Full detail in docs/engineering.md.

Stage a result from THIS repo into the Noumenal Verified ML Journal. Use when a contributor wants to submit, stage, or publish a Lean result (a theorem + optional executable witness) to the journal / the Noumenal site, or asks "how do I add my repo / my theorem to Noumenal", "stage this for the journal", "make a noumenal.json", or "submit this unit". Produces a verified noumenal.json (schema unit.v2) and opts the repo in. Honesty-enforcing: it derives the tier and the trust base from the real artifacts and refuses to stamp an unproven result as accepted.