Shilpi 1.7, released in August 2026, moves the AI workflow beyond the visual canvas. Earlier releases made pages, templates, variables and project assets available to an agent. Version 1.7 adds the surrounding operations needed to finish and distribute a real WordPress site.
34 scoped MCP tools
The WordPress-hosted MCP server now exposes 34 tools, up from 30 in version 1.6. The new abilities cover SEO, WooCommerce, menus, taxonomies and whole-site import. They remain bounded tools rather than general access to the server or theme files.
SEO metadata through the active plugin
An agent can read and write the SEO title, meta description, canonical URL, robots directives and social-card fields for a page or post. Shilpi detects supported SEO plugins and uses their own APIs, so existing Yoast, Rank Math or All in One SEO data stays inside the system that owns it. Plugin caches and sitemaps can remain consistent.
WooCommerce becomes addressable
The same connection that builds pages can now inspect and configure a WooCommerce storefront, manage simple products and author product routes. WooCommerce still owns cart, checkout, stock, gateways and customer accounts. The MCP surface does not expose customers, orders, addresses or payments.
Menus and taxonomies are first-class operations
An agent can create and order navigation items, assign a menu to Primary or Footer, create categories and tags, and connect taxonomy terms to posts. Query loops can then target those terms without a separate manual pass through wp-admin.
Three new template types
Search, 404 and Loop join Header, Footer, Single, Archive and Part, bringing the total to eight template types. Template assignment continues to follow WordPress-style slugs. A reusable Loop template can power blog archives, search results and related-post sections from one card design.
Site manifests and whole-site import
A site manifest can describe pages, templates, parts and the design system as one document. The whole-site importer brings an HTML project into WordPress in a single operation, extracts byte-identical headers and footers, downloads images, converts reusable styles and builds navigation. Repeating an import updates stable slugs instead of duplicating the site.
Managed script files and load order
Trusted JavaScript files can be uploaded, kept disabled until reviewed and assigned an explicit load priority. Dependencies can load before plugins and page behaviour, while jQuery-based libraries can request WordPress’s registered jQuery rather than assuming it exists.
Compatibility update
Shilpi 1.7 was tested against WordPress 7.1 and PHP 8.3. The release continues the project’s approach of listing fixes and constraints alongside new capabilities.
What this changes for agencies and site owners
An agent can now move through a more complete delivery loop: inspect the existing site, build pages and templates, organize content, configure metadata, prepare the storefront, verify saved records and reproduce the system elsewhere. The value is not simply more automation; it is fewer untracked steps between design and a finished WordPress website.
Frequently asked questions
What is new in Shilpi 1.7?
The headline additions are expanded MCP tools, SEO metadata management, WooCommerce support, menu and taxonomy operations, Search/404/Loop templates, site manifests, whole-site import and managed script load order.
Does Shilpi 1.7 require an SEO plugin?
Its SEO tools integrate with supported active SEO plugins rather than writing hidden raw metadata. Check the current documentation for supported drivers.
Can a whole HTML site be imported into WordPress?
Yes. Shilpi can import multiple HTML pages, identify shared header and footer markup, bring images into the Media Library and create stable WordPress page records.
Read the complete release history
See the official Shilpi changelog, explore every feature, or open the documentation for implementation details.