Praxis

Praxis is the Methodological Context / Know-how component of the MOLI Platform.

Praxis represents reusable scientific know-how independently of any single DiscoveryProject or modeling implementation.

Its central distinction is:

Capability = WHAT we know how to do.
Protocol = HOW we reproducibly do it.

Role in MOLI

Scientific Context
       │
       ├── Sabueso  — Knowledge
       ├── Praxis   — Know-how
       └── Nextia   — Discovery

A Capability describes a semantic scientific ability, its applicability, contracts, validation, limitations, and available implementations.

A Protocol is a reproducible implementation of a Capability and may orchestrate MolSysSuite components and external scientific engines.

Praxis does not replace scientific APIs and must not hide expert access to them.

Methodological learning

scientific need
      ↓
APIs / tools
      ↓
ad-hoc workflow
      ↓
formalized Protocol
      ↓
validation / generalization
      ↓
Capability

Successful project execution does not automatically create validated methodology. Promotion into reusable Praxis know-how is explicit and gated.

Status

Praxis now includes an experimental local Python implementation: a versioned catalog with explicit admission/assessment/suitability history, recorded selection policies, typed quantity contracts, bounded composition and human gates, method records, physical benchmarks/reanalysis, audit, replay preflight and Markdown/HTML reports. APIs and storage formats are not frozen.

Version 0.1.0 is the frozen first experimental local-methodology source milestone, tagged at e253fe7. The release record identifies its passing gates, qualified artifacts and known limitations. Exact-source acceptance is recorded in Praxis #9; configured CI alone is not passing evidence. No public Conda package or archival claim is made.

The public source tag 0.1.1 at 2a5149b corrects runtime-contract and human-review audit defects and receives newer public support providers. Its accepted release record and tracked evidence preserve scope, passing gates and original hashes under Praxis #9. Package publication remains separate. Development is paused as of 2026-10-26; the pause checkpoint records how to resume.

The first implementation guide explains the boundaries and the optional Recorda and Ackredit integrations. Bundled hydrogen-refinement definitions retain their pending adapters and checks; no executable hydrogen engine or scientific validation is claimed.

For development, provision devtools/conda-envs/development_env.yaml and install this checkout with python -m pip install --no-deps --no-build-isolation --editable .. Run python examples/local_comparison.py /tmp/praxis-example for the fictional paired-coordinate fixture. With MolSysMT provisioned, run python examples/molsysmt_comparison.py /tmp/praxis-molecular LOCAL_PDB for a real C-alpha RMSD exercise, physical metrics and reports. These do not certify methods or rank fitted and fixed-frame observables as interchangeable.

The remaining work identifies shared-boundary, hydrogen-provider and publication owners. The earlier implementation checkpoint is historical. The qualified noarch Conda artifact is retained without upload; the public source tag does not provide a public package installation route.

Python metadata admits 3.11–3.14. Local development uses Python 3.14. Qualified releases of the experimental local package support Linux x86_64 and macOS arm64; their same-artifact installed matrix and engineering adoption are recorded in Praxis #9. Windows is unclaimed. macOS support is currently limited to Apple Silicon (arm64). Intel-based macOS (x86_64) is not part of the supported platform matrix. Support may be reconsidered if there is demonstrated user demand. Praxis has no published installation route yet.

Coverage measures the full installed Conda suite on Linux Python 3.14 with published Recorda/Ackredit providers; the optional MolSysMT example is outside this scope. Reports upload on main pushes, manual runs and weekly. The first accepted report measured fc51272 on 2026-10-08; subsequent report identities are recorded in #9. The live percentage describes this test scope, not scientific validation.

The package is released under the MIT license.

First drafts of the Capability and Protocol proposal templates provide a starting point for describing and reviewing experimental methodology.

The initial programming design records the proposed package structure and open decisions for discussion.

The normative conceptual definition is maintained in MOLI Platform Architecture 1.0.

Initial design

See the Initial methodological slice proposal and implementation issue.