Interactivity inference: AST analysis for hooks + event handlers

[qantrepreneur]

What to build

Implement the static analysis module that determines whether a component (and its import tree) is interactive. This is the Marko-style auto-inference engine.

The analyzer receives a component's source code and its resolved import graph, then checks for signals of interactivity:

• React hooks: useState, useEffect, useRef, useReducer, useCallback, useMemo, useSyncExternalStore, useTransition
• Event handler props in JSX: onClick, onChange, onSubmit, onKeyDown, etc.
• Browser APIs: window, document, addEventListener, navigator
• Custom hooks (functions starting with use that call other hooks)

If a component OR any component in its import tree uses any of these, it's marked interactive.

Returns a boolean per component/slot.

Acceptance criteria

• Detects React hook usage in a component
• Detects event handler props in JSX
• Detects browser API usage
• Walks the import tree (transitive dependencies)
• A static component importing only static children → not interactive
• A static component importing one interactive child → interactive
• Pure function: accepts source code + import graph, returns boolean
• Unit tests for each detection category and transitive inference

Blocked by

• Core types: Context, SlotMap, TemplateRegistry, RouteMap #2