Platform Support & Limits

AgentShelf supports 7+ platforms with different levels of access and capabilities.

WordPress

Detection: WP REST endpoint presence, WordPress meta tags, HTML patterns

What We Read

WP REST API (/wp-json/wp/v2/posts) — when public
HTML fallback — post list parsing when REST is locked
Post meta — titles, excerpts, featured images, dates
Open Graph — og:title, og:description, og:image

WooCommerce

If WooCommerce is detected:

Product loop HTML parsing
WooCommerce-specific structured data
Product cards and pricing

Limits

No authentication support (public only)
REST API must allow anonymous reads (default in most WP installs)
HTML fallback is best-effort for locked sites

Recommendations

Enable WP REST API for public reads (most WP sites already do this)
Install Yoast SEO or similar for proper Open Graph tags
Use structured data plugins (Schema, Rank Math) for product markup

Official Plugin (Planned)

A WordPress plugin for one-click llms.txt publishing is on the roadmap.

Shopify

Detection: Shopify meta tags, /products.json endpoint

What We Read

/products.json — full product catalog (public API)
/collections.json — collection list for llms.txt sections
Product data — title, description, variants, images, vendor

Limits

Public catalog only — no Admin API access
Products must be published and visible online
Does not read Shopify apps or custom fields not in products.json
Pagination: fetches first 250 products by default

Recommendations

Publish all products you want AI agents to see
Use descriptive product titles (avoid SKU-only names)
Write full descriptions — Shopify's description field is fully readable
Tag products properly — helps with llms.txt organization

Official Plugin (Planned)

Shopify app for automatic llms.txt sync and bot analytics coming in Pro.

Etsy

Detection: Etsy shop URLs, HTML patterns

What We Read

Shop page HTML — public listing cards
Listing data — title, price, image, link
Shop meta — shop name, description

Limits

HTML parsing only — no official Etsy API integration yet
Shop must be public
Best-effort extraction (Etsy may change HTML structure)
OAuth support planned for future releases

Recommendations

Write clear listing titles with keywords
Use high-quality images (helps with llms.txt previews)
Fill out shop announcement and description for better meta scoring

Official Integration (Planned)

OAuth + Etsy API integration planned for Pro.

WooCommerce

Detection: WooCommerce JavaScript, product schema, HTML patterns

What We Read

WordPress data (see WordPress above)
WooCommerce product loops and card markup
Product schema (JSON-LD structured data)
Pricing and availability

Limits

Same as WordPress (public only)
Product data quality depends on theme and plugins

Recommendations

Use a modern WooCommerce theme with proper structured data
Install WooCommerce SEO or Yoast for optimal meta tags
Enable product schema output (most themes do this by default)

Strapi

Detection: Strapi meta tags, /api endpoint probing

What We Read

Public /api/<content-type> endpoints
Entry collections — products, articles, etc.
Attributes — titles, descriptions, images, relations

Limits

Public API only by default
If Strapi Public role is locked, you must provide STRAPI_API_TOKEN via MCP env
Pagination: fetches first 100 entries per content type
Requires Strapi v4+ (Strapi v3 not officially supported)

Credentials Setup (Optional)

If your Strapi API requires authentication:

Generate an API token in Strapi admin (Settings → API Tokens)
Set STRAPI_API_TOKEN in your MCP config:

{
  "mcpServers": {
    "agentshelf": {
      "env": {
        "STRAPI_API_TOKEN": "your-token-here"
      }
    }
  }
}

Recommendations

Enable Public API access for AgentShelf to read (or provide token)
Use descriptive field names (title, description, image)
Populate SEO fields if using Strapi SEO plugin

Contentful

Detection: Contentful CDN URLs, page metadata

What We Read

Contentful Delivery API — requires space ID and access token
Entry collections — products, articles, pages
Assets — images and files

Limits

Credentials required (Contentful has no public API)
Must provide CONTENTFUL_SPACE_ID and CONTENTFUL_ACCESS_TOKEN
Fetches first 100 entries per content type
Requires Content Delivery API access (not Preview API)

Credentials Setup

Get your Space ID from Contentful settings
Generate a Content Delivery API token
Set environment variables in MCP config:

{
  "mcpServers": {
    "agentshelf": {
      "env": {
        "CONTENTFUL_SPACE_ID": "abc123xyz",
        "CONTENTFUL_ACCESS_TOKEN": "your-cda-token"
      }
    }
  }
}

Recommendations

Use a dedicated CDA token for AgentShelf (read-only)
Structure content models with title, description, and image fields
Publish all content you want indexed (unpublished entries are not fetched)

Generic / Fallback

Detection: Any public website

What We Read

JSON-LD structured data (Product, Offer, Organization)
Open Graph tags (og:title, og:description, og:image)
Semantic HTML — headings, lists, product cards
Microdata (schema.org attributes)

Limits

Best-effort parsing only
Quality depends on site markup
May not recognize all product patterns

Recommendations

Add JSON-LD structured data for products (schema.org/Product)
Use Open Graph tags for social/meta previews
Semantic HTML — proper headings (H1, H2, H3), lists, and links

Platform Detection Priority

AgentShelf detects platforms in this order:

Shopify (strongest signals)
Etsy (shop URL pattern)
Strapi (API endpoint presence)
Contentful (CDN detection)
WooCommerce (WC-specific scripts/schema)
WordPress (WP REST or meta tags)
Generic fallback (all other sites)

If multiple platforms match, the most specific one wins.

Need more details? Check the User Guide or MCP Setup.