Closure 0.1.0a1 — early release candidate

Closure connects engineering requirements, inputs, analyses, and results in readable project files. Change an input, run Verify, and follow the affected work through to the evidence behind each requirement.

This early release candidate brings together the Closure 0.1 specification, Python CLI/backend, and local browser viewer. It is available under Apache-2.0 from Keystone Intelligence LLC.

Get started

Use Python 3.10 or 3.12 on Windows or Linux. Download closure_se-0.1.0a1-py3-none-any.whl below, create and activate a virtual environment, then run:

python -m pip install ./closure_se-0.1.0a1-py3-none-any.whl
closure --version
closure init -p my-project
cd my-project
closure check

Follow the quickstart to review the starter analysis, approve its baseline, run Verify, and open the viewer. The viewer is included in the wheel; Node is needed only for frontend development and JavaScript examples.

• Specification and runner protocol
• CLI guide
• Viewer guide
• Compatibility and preview scope

What is included

The source archive includes versioned schemas, conformance fixtures, examples across engineering domains, and the implementation. The wheel and source distribution are the exact artifacts tested by the Linux/Python 3.12 CI job. source-manifest.json lists the public source files, release-record.json identifies the release commit and CI run, and SHA256SUMS.txt covers the release assets.

The complete Windows/Linux × Python 3.10/3.12 matrix covers core behavior and fresh installations, with Chromium and Firefox workflows on Python 3.12. Optional CAD and external solver integrations are outside that matrix.

Help shape the next release

Try an analysis you already use. Tell us where setup gets in the way, which fields feel unnecessary, and whether you can explain a changed result. Open an issue with your feedback. Independent first-session onboarding is deliberately being collected during this preview; no completed independent usability study is claimed.

Analyses execute as trusted local code. Verify assesses the declared model and dependencies; it does not certify a physical design. See the compatibility notes for the Python numeric digest profile and other preview boundaries. Report security issues through GitHub private vulnerability reporting.