Linux 软件免费装
Banner图

KaliCart MCP

开发者 carthub
更新时间 2026年6月25日 16:08
PHP版本: 8.0 及以上
WordPress版本: 7.0
版权: GPLv2 or later
版权网址: 版权信息

标签

ai chatgpt ai search claude mcp

下载

0.2.10 0.2.11

详情介绍:

Help AI assistants find, read and cite your WordPress content. KaliCart MCP makes your WordPress site agent-ready: it turns your editorial content into something AI agents and assistants — such as ChatGPT, Claude, Gemini, Perplexity and any MCP-capable client — can browse, search and read directly. It exposes your posts, pages and public custom post types as a standards-based Model Context Protocol (MCP) server over JSON-RPC 2.0, alongside clean Markdown output and a lightweight presence/discovery layer. No LLM, no cloud, no external calls — everything runs on your own server. It is read-only: an agent connects to a single endpoint on your site and can list, search, and read your published content as structured data — it never creates, edits, or deletes anything. What it does Content stays under your control Exclusion works on three levels, all enforced:
  1. Structural - attachments and commerce objects (products, variations) are never exposed. MCP is for content, not commerce.
  2. WooCommerce functional pages - cart, checkout, my account, shop, and policy pages are excluded automatically as application UI rather than editorial content.
  3. Per-item - a "Hide from AI agents" checkbox in the editor lets you remove any single post or page.
You choose which post types are exposed from a simple toggle in the admin screen. WooCommerce coexistence KaliCart MCP is the content companion to the KaliCart Bridge commerce plugin. The two are fully independent - separate namespaces, separate code - and can run side by side. MCP handles editorial content; Bridge handles products and commerce.

安装:

  1. Upload the plugin files to /wp-content/plugins/kalicart-mcp, or install through the Plugins screen in WordPress.
  2. Activate the plugin through the Plugins screen.
  3. Open the KaliCart MCP menu to see your endpoint, choose which content types to expose, and copy the ready-to-paste Claude Desktop configuration.

升级注意事项:

0.2.11 Richer agent navigation: taxonomy terms with slugs in site_info, publish-date filtering, word counts, and an optional author toggle. 0.2.5 New admin header logo. 0.2.4 Restores the toggle switch appearance. 0.2.3 Completes the toggle click-area fix for the content exposure section. 0.2.2 Fixes the exclusion toggle reacting to clicks on the whole row. 0.2.1 Minor admin UI fixes (menu icon and toggle behavior). 0.2.0 Adds content exclusion controls, automatic WooCommerce page exclusion, and IT/DE/FR/ES translations. 0.1.0 Initial release.

常见问题:

How is this different from the official WordPress MCP Adapter?

They do opposite jobs, and a site can use both. The official MCP Adapter (built into WordPress 7.0) lets an AI agent manage your site for you — create drafts, update posts, moderate comments — after you log in and authorize it through a connector. It is built for the site owner operating their own site through an AI assistant. KaliCart MCP needs no connector and no login. You install it, and it makes your published content discoverable and readable by AI assistants and agentic search — the way search engine indexing made your site findable on Google. It manages the surface of data your site exposes to AI: ChatGPT, Claude, Gemini, Perplexity and any agent can find, read and cite your pages. It only ever reads, and only your editorial content — never your admin, settings, or commerce data. In short: the official adapter is the control panel an agent uses to run your site; KaliCart MCP is the index an agent reads to answer questions about your site — the SERP for agentic search. Many sites will want both.

How do I make my WordPress site readable by AI assistants like ChatGPT and Claude?

Install KaliCart MCP and activate it. It exposes your posts and pages through a Model Context Protocol (MCP) server — the standard interface AI assistants use to read external content. Open the KaliCart MCP admin screen, choose which content types to expose, and connect any MCP-capable client using the ready-to-paste configuration.

What is an MCP server for WordPress?

MCP (Model Context Protocol) is an open standard that lets AI assistants connect to external data sources. An MCP server for WordPress turns your site into a source that assistants like Claude, ChatGPT, Gemini and Perplexity can browse, search and read directly, instead of relying on scraping or guesswork.

Does it work with Claude, ChatGPT, Gemini and Perplexity?

Yes. KaliCart MCP implements the standard Model Context Protocol over JSON-RPC 2.0, so any MCP-compatible client or assistant can connect. The admin screen includes a copy-paste configuration for Claude Desktop using mcp-remote.

Will this help my content appear in AI search and AI answers?

It makes your content accessible to AI agents in a clean, structured form, which is the prerequisite for an assistant to read and cite it accurately. The plugin makes your content readable; it does not promise rankings or placement, which no plugin can guarantee.

Does this send my content to any external service?

No. The plugin makes no outbound calls. It only responds to agents that connect to your own endpoint. There is no LLM and no cloud component.

Is it read-only?

Yes. Agents can read, list, and search your published content. The plugin never creates, edits, or deletes anything.

How do I connect an agent?

The admin screen provides a copy-paste Claude Desktop configuration using mcp-remote. Any MCP-compatible client can connect to the endpoint at /wp-json/kalicart-mcp/v1/mcp.

Will my products show up?

No. Products, variations, attachments, and WooCommerce functional pages are excluded automatically. MCP exposes editorial content only.

Can I hide a specific page?

Yes. Each post and page has a "Hide from AI agents" checkbox in the editor sidebar.

Do I need WooCommerce?

No. WooCommerce is optional. If it is installed, the relevant functional pages are detected and excluded automatically.

更新日志:

0.2.11 0.2.10 0.2.9 0.2.8 0.2.7 0.2.6 0.2.5 0.2.4 0.2.3 0.2.2 0.2.1 0.2.0 0.1.0