Modules

New in v3.1.0. Modules are substantial features you turn on and off from EMCP Tools → Modules. Unlike individual tool toggles, a module owns a whole feature, its admin surface, hooks, cron, and (where relevant) MCP tools, so disabling it is a true kill switch.

Each module boots only when active. Active module IDs live in a single option; each module keeps its own settings separately.

Available modules

Module	Tier	Default	What it does
Image Optimization	Free	Off (opt-in)	Compress uploads + generate/serve WebP
EMCP Themer	Free	On	Builder-agnostic theme builder
Prompts	Free	On	The prompt library tab
Brand Kits	Free	On	Curated color + typography kits
Templates	Pro	On	The premium template library
AI Chat	Pro	On	In-editor AI chat (Elementor + Gutenberg)
Agent Skills	Pro	On	Runtime skill exposure: the list-skills/get-skill Agent Skills tools + discovery-context catalog

A module with a dedicated screen shows a Configure → link on its card; a lighter module shows an inline settings overlay.

Image Optimization

Free and opt-in (it mutates your uploads, so it ships off). When enabled it:

• Compresses generated image sub-sizes on upload and generates .webp siblings, using WordPress’s own WP_Image_Editor, with no external binaries.
• Serves WebP by rewriting wp_get_attachment_* URLs. On the front end this is gated by the browser’s Accept: image/webp header; in REST and WP-CLI it’s always served, so the MCP media tools (get-media, sideload-image, add-stock-image, …) resolve to WebP with no tool changes.
• Preserves full-size originals (optionally trimmed by a max-dimension cap) and backs up every touched file under uploads/emcp-originals/, so the whole thing is reversible.
• Includes a resumable bulk optimizer (processes your existing library in batches) and a Restore originals companion.
• Adds an in-place image resizer and the emcp-tools/resize-media MCP tool (new in v3.1.1). An AI agent can resize an existing Media Library image by ID, mid-build or after the fact.

Settings: compress on/off, WebP generation on/off, front-end WebP serving on/off (REST/CLI always serve WebP), quality (1–100, default 60), max dimension (0 = off), keep originals. Availability depends on your server’s image library supporting WebP output. The card disables the toggle and explains when it can’t.

emcp-tools/resize-media

Resizes an existing Media Library attachment in place (added in v3.1.1). By default it scales the image to fit within the given width/height (aspect ratio preserved); pass crop: true for an exact width × height. The original is backed up under uploads/emcp-originals/ (reversible), all sub-sizes + WebP siblings are regenerated, and the attachment ID and URLs stay the same, so anything already pointing at the image keeps working. Registers only when the Image Optimization module is enabled.

Input	Required	Notes
attachment_id	yes	The Media Library image to resize.
width	one of width/height	Target width in px (≥ 1).
height	one of width/height	Target height in px (≥ 1).
crop	no	true = crop to an exact width × height (both required); default scales to fit.

Agent Skills

Pro and on by default. This module is the on/off switch for runtime skill exposure. It registers the read-only list-skills and get-skill tools and injects the ## Skills catalog into the MCP discovery context, so a connected agent can browse and load the bundled playbooks over the MCP connection. Turning it off removes those tools and the discovery injection (reclaiming their token footprint); it does not affect the local Skills download on the Skills tab. Full detail in Agent Skills.

AI Chat

Pro and on by default. Owns the whole AI Chat feature: the AI Chat admin tab, the REST routes, the conversation store, and the floating chat panel in both page editors, Elementor and the WordPress block editor. Turning the module off is a true kill switch: the tab, the editor panels, the routes, and the cron all stop.

Bring your own API key for Anthropic, OpenAI, OpenRouter, or Google Gemini. Keys are encrypted per user and never leave your site except to the provider you chose.

Reading a web page: web_fetch

New in v3.2.0. The chat’s model can read a URL you give it: a reference design, a competitor page, documentation, a JSON endpoint, or a stylesheet.

• format: "text" (the default) returns the page’s readable text, cheap, and usually all you need.
• format: "html" returns sanitized markup including the <head>, when the model needs class names to reason about layout.
• Fetching an HTML page also returns a stylesheets array of absolute CSS URLs. Ask the AI to read one, and it lifts the design’s color tokens and type scale straight from source, far cheaper and more accurate than inferring them from a rendered page.

It cannot search the web: you have to give it a URL. Only text is readable (HTML, CSS, JSON, XML, Markdown, plain text). Images, fonts, and scripts are refused.

It fetches from your server

The chat runs in your browser, but a browser cannot read a third-party page (CORS), so the request is made by your WordPress server. Every URL (and every redirect hop, followed manually rather than by the HTTP client) is validated before a request goes out: http(s) only, no credentials in the URL, ports 80 and 443 only, and refused if any address the host resolves to is loopback, private, link-local (including the cloud metadata address 169.254.169.254), CGNAT, multicast, or reserved.

Responses are capped at 2 MB, restricted to textual content types, and never executed. Fetched content reaches the model wrapped and labelled as untrusted data, and the model is instructed never to act on instructions found inside a fetched page.

A practical consequence: you cannot point web_fetch at your own site when it resolves to a private address, a local .test domain on 127.0.0.1, or a staging box on 10.x. That is the protection working as intended.

Building your own

The framework is extensible: Pro units register modules through the plugin’s loader, and the emcp_tools_register_modules action lets code register a custom module. See Filters & hooks.