An MIT-licensed static prototype for turning plain-language situation inputs into a transparent, structured comparison of competing explanations.
NSENT uses deterministic lexical matching to rank explanations supplied by the user. Its normalized support scores are not real-world probabilities and do not establish causation, intent, or wrongdoing. See METHODOLOGY.md for the scoring model and limitations.
No build step is required. Open index.html directly or serve this directory with any static server.
python -m http.server 8080Then open http://localhost:8080. No package installation, build step, account, API key, or backend is required.
The source is public and can run locally. The Retehost-hosted edition uses a separate sign-in gate for founder-approved accounts; that hosting configuration is maintained in the private Retehost repository. Public source access does not grant access to the hosted workspace.
This edition processes form entries locally in your browser. It includes no analytics, external requests, or saved analysis storage. Reloading the page does not restore a saved analysis.
index.html— semantic application shellassets/styles.css— complete responsive design/brand systemassets/app.js— form state, dynamic fields, live formula, renderingassets/engine.js— isolated deterministic analysis engineMETHODOLOGY.md— formula, interpretation, limitationsHANDOFF.md— product and engineering handoff
The engine is deliberately isolated from the UI. A future LLM/NLP extraction adapter can replace engine.js without redesigning the application.
Issues and pull requests are welcome. Keep changes focused and explain how you verified them. Preserve the distinction between lexical support scores and real-world probabilities.
Copyright (c) 2026 Clear Frameworks. Released under the MIT License.