engineering-prototype-kit

Organize contracts, proposals, meeting notes, screenshots, interface notes, and verbal requirements into confirmed facts, goals, roles, core objects, flows, assumptions, open questions, dependencies, and risks. Use before planning a demo, preparing a business interview, or when source materials conflict and must not be silently turned into requirements.

Register and process change requests through explicit authorization, readiness, human-intervention, evidence, and delivery gates. Use when a user says record, plan, start, implement, review, synchronize, unblock, or deliver a CR; asks whether AI may execute; or needs typed artifacts and validation without silently closing or freezing work.

Classify review feedback into positive decisions, negative decisions, visual preferences, unresolved business questions, and technical questions, then model core domain objects and testable state machines after confirmation. Use when feedback changes product outcomes, multiple pages disagree on state meaning, or a rule must be protected from accidental restoration.

Create or update structured page contracts and readable page specifications from confirmed decisions, state machines, and a runnable prototype. Use when formal pages need stable Page, Action, and State IDs; when prototype behavior must become testable acceptance facts; or when page documentation must be synchronized without inventing APIs or restoring removed fields.

Inspect and guide an AI-assisted engineering prototype project through gated stages while classifying execution state, AI policy, readiness, human intervention, and completion artifacts. Use when a product manager asks where to start, what can run now, why work is blocked, what to do next, to continue an existing project, or to coordinate prototype delivery without crossing approval gates.

Run a read-only audit across task control, decisions, state machines, page contracts, runnable prototypes, page specifications, readiness evidence, and typed artifacts. Use when checking structural completeness, authorization, human-intervention packets, consistency, removed-item residue, review readiness, or P0/P1/P2 evidence before fixes.