Query & Discovery

These tools never mutate state. They require edit_posts and are safe to call before any write operation. They’re how the AI figures out what exists before changing anything.

`emcp-tools/list-widgets`

Returns every registered widget type with names, titles, icons, categories, and keywords. The AI uses this to discover which widgets are available (free + Pro + custom + third-party).

Returns: array of { name, title, icon, categories, keywords }.

`emcp-tools/get-widget-schema`

Returns the curated params for a widget by default: the settings you actually need, with types and defaults. Pass a types array to inspect several widgets at once, or full: true for the complete auto-generated control schema. Used before add-free-widget / add-pro-widget to know which settings are valid.

Input: widget_type (e.g. "heading", "button", "form"). Returns: curated params object (or the full JSON Schema when full: true).

`emcp-tools/get-page-structure`

Returns the full element tree for an Elementor-enabled page: containers, widgets, nesting, IDs. The starting point for any “edit this page” workflow.

Input: post_id. Returns: tree of { id, elType, widgetType?, settings, elements: [...] }.

`emcp-tools/get-element-settings`

Returns the current settings of a specific element by ID.

Input: post_id, element_id. Returns: settings object.

`emcp-tools/list-pages`

Lists every Elementor-enabled page or post on the site.

Returns: array of { id, title, status, post_type, edit_url }.

`emcp-tools/list-templates`

Lists saved templates from the Elementor template library.

Returns: array of { id, title, type, date, author }.

`emcp-tools/get-global-settings`

Returns the active Elementor kit’s global settings: color palette, typography presets, spacing, default fonts.

Returns: kit settings object.

Typical flow

1. list-pages          → find the target page
2. get-page-structure  → understand existing layout
3. get-element-settings → inspect a specific element you want to modify
4. update-widget       → apply changes

Reading before writing is the most reliable pattern for the AI. It prevents the “regenerate from scratch” failure mode and keeps existing content intact.