Communication Preflight Prototype v0.1

Status: Research-only executable reference model Date: 2026-09-20

Purpose

Test the smallest deterministic core implied by Operational Context Experiments 04-05.

The prototype intentionally does not parse natural-language requests or infer audience, authority, culture, or purpose.

Instead, it assumes a human/model has produced a typed communication-state assessment with:

• field identity;
• state: known / unknown / not-applicable;
• provenance;
• materiality;
• clarification group;
• whether a safe bounded default exists.

The prototype then mechanically derives:

• material missing fields;
• optional unknown fields;
• grouped clarifications;
• activation decision;
• whether bounded progress is permitted;
• verification escalation.

Why this boundary matters

Natural-language interpretation is a Decide / sometimes Deliberate problem.

Once the semantic assessment exists, the remaining workflow can largely be Compute.

That prevents the runtime from hiding audience or authority guesses inside deterministic-looking tooling.

Candidate outcomes

• DraftNormally
• PreflightNeeded
• ClarificationRequired
• HighConsequenceVerificationRequired
• AbstainForAuthorityOrPermission

Files

• fixtures.json — semantic assessments corresponding to Experiments 01, 04, and 05.
• reference-preflight.mjs — executable research reference.
• expected-results.json — frozen expected results.
• PreflightSketch.fsx — F# design sketch for a future typed implementation.
• test-preflight.mjs — verifies deterministic behavior.

Non-goal

This prototype is not the product.

It does not:

• generate prose;
• infer demographic preferences;
• decide cultural permission;
• score communication quality;
• replace user clarification;
• implement an npm lifecycle tool.

If promoted, production implementation should follow the project's normal F#/typed-state architecture.