ars-nova-wordpress-mcp

Local MCP server for managing arsnovasingers.org via the WordPress REST API.

Built for Jonathan Raabe (Marketing Director, Ars Nova Singers) so Claude can read, audit, and (in later phases) update the WordPress site directly.

Phases

Phase	Scope	Status
1	Read-only audit (posts, pages, media, users, plugins, themes, settings, SEO meta)	In progress
2	Content writes (create/update posts + pages, upload media)	Planned
2.5	WPEngine API integration (cache flush, perf metrics, backups)	Deferred until needed
3	Custom Ars Nova endpoints (events, season pages, ACF fields if discovered)	Planned
4	Eval suite (10 read-only complex questions)	Planned

Setup

1. Generate a WordPress Application Password

1. Log into wp-admin at https://arsnovasingers.org/wp-admin
2. Go to Users → Profile
3. Scroll to Application Passwords
4. Name it claude-mcp and click Add New Application Password
5. Copy the 24-character password — WordPress only shows it once

2. Install + build

cd ars-nova-wordpress-mcp
npm install
npm run build

This produces dist/index.js which is the entry point Claude Desktop runs.

3. Add to claude_desktop_config.json

{
  "mcpServers": {
    "ars-nova-wordpress": {
      "command": "node",
      "args": [
        "C:\\Users\\jonra\\Claud Projects\\Ars Nova\\Ars Nova\\ars-nova-wordpress-mcp\\dist\\index.js"
      ],
      "env": {
        "WP_SITE_URL": "https://arsnovasingers.org",
        "WP_USERNAME": "your_wp_username",
        "WP_APP_PASSWORD": "your_24_char_password"
      }
    }
  }
}

Restart Claude Desktop after editing.

4. Smoke test

In a Claude Desktop conversation, ask:

Use the ars-nova-wordpress MCP to call wp_get_site_info.

If it returns the site name, tagline, and WP version, auth is working.

Tool list (Phase 1)

Content

• wp_list_posts — list posts with filters
• wp_get_post — fetch one post by ID or slug
• wp_list_pages — list pages with filters
• wp_get_page — fetch one page by ID or slug
• wp_search_content — search across posts and pages

Media

• wp_list_media — list media library items
• wp_get_media_item — fetch one media item with full metadata

Users

• wp_list_users — list users and their roles
• wp_get_user — fetch one user by ID

Taxonomy

• wp_list_categories — list categories
• wp_list_tags — list tags

Site

• wp_get_site_info — site name, tagline, WP version, REST namespaces
• wp_list_themes — installed themes (active flagged)
• wp_list_plugins — installed plugins (active flagged, version reported)
• wp_get_settings — general WP settings
• wp_list_post_types — registered post types (reveals custom post types)

SEO

• wp_get_seo_meta — auto-detects Yoast / RankMath / AIOSEO and returns the per-page SEO meta
• wp_audit_alt_text — scans the media library for items missing alt text

Tool conventions

Every tool:

• Uses snake_case naming with wp_ prefix
• Validates inputs with Zod (.strict())
• Supports response_format: "markdown" | "json" (default markdown)
• Implements pagination with limit, offset, has_more, next_offset
• Returns both content[] (text) and structuredContent (typed object)
• Declares MCP annotations: readOnlyHint, destructiveHint, idempotentHint, openWorldHint
• Truncates responses larger than CHARACTER_LIMIT (25,000) with a clear message
• Surfaces 401/403/429 errors with actionable next-step guidance

Auth model

WordPress Application Passwords (built into WP 5.6+, no plugin required). Sent as Basic Auth on every request. Application Passwords can be revoked individually from the WP profile page without affecting the user's normal login.

Project structure

ars-nova-wordpress-mcp/
├── package.json
├── tsconfig.json
├── README.md
├── src/
│   ├── index.ts             # Entry point — registers all tools, starts stdio transport
│   ├── constants.ts         # CHARACTER_LIMIT, default limits, API namespace
│   ├── types.ts             # Shared TypeScript interfaces
│   ├── services/
│   │   └── wp-client.ts     # Axios client with auth + shared error handling
│   ├── schemas/
│   │   └── common.ts        # Shared Zod schemas (pagination, response_format)
│   └── tools/
│       ├── content.ts       # Posts, pages, search
│       ├── media.ts         # Media library
│       ├── users.ts         # Users, roles
│       ├── taxonomy.ts      # Categories, tags
│       ├── site.ts          # Site info, themes, plugins, settings, post types
│       └── seo.ts           # SEO meta + alt text audit
└── dist/                    # Build output — entry point: dist/index.js