# Advanced MCP Plugin — Overview

## What It Is

A custom WordPress plugin that exposes a full **Model Context Protocol (MCP) server** inside WordPress. It lets Claude (via the claude.ai MCP connector or any MCP client) directly read and write WordPress content through a secure, scoped HTTP API — no WP-CLI, no SSH, no manual file editing required.

## What It Does

Claude connects to the plugin's MCP endpoint and gains hands-on access to:

- Create, edit, and publish posts, pages, and custom post types
- Build complete Elementor layouts programmatically (containers, headings, text, buttons, widgets)
- Manage media, users, comments, menus, taxonomies
- Read and write WooCommerce products, orders, and customers
- Update SEO meta (Yoast / RankMath)
- Run WP-CLI commands through a sandboxed allowlist
- Execute SELECT-only raw SQL queries
- Manage files in the uploads directory
- Read theme settings, active Elementor kit tokens, WoodMart widgets

## Installed Environment

| Item | Value |
|---|---|
| Host | mainwp.asktam.net |
| Container | `mainwp` (Docker) |
| VPS | 76.13.157.7 |
| Plugin path | `/var/www/html/wp-content/plugins/advanced-mcp-plugin/` |
| DB prefix | `mwp_` |
| Theme | WoodMart (Elementor + WooCommerce) |
| PHP | 8.x |
| WP-CLI | `/usr/local/bin/wp` (requires `--allow-root`) |
| WP Admin credentials | mainwp / sameerismE2!! |

## Repo

- **Gitea:** `git.asktam.net/colorlink/advanced-mcp-plugin`
- **Branch:** `main`
- **Author:** colorlink <loveriver7@gmail.com>

## Version

`2.1.0` — security hardened, Grade A+ audit (30/30 checks passed 2026-05-31)

## MCP Endpoint

```
POST https://mainwp.asktam.net/wp-json/asktam-mcp/v1/mcp
Authorization: Bearer <jwt_token>
Content-Type: application/json
```

Supported MCP methods:
- `initialize` — server info + capabilities
- `tools/list` — all available tools with input schemas
- `tools/call` — execute a tool

Protocol: JSON-RPC 2.0 over HTTP.

## Tool Count

**123 tools** across 18 tool groups. See `docs/03-tools.md` for the full catalogue.

## Security Grade

A+ (30/30 security checks). See `docs/04-security.md` for full audit results.