Schema & Meta Generation with OpenClaw

## meta-writer

When asked to generate meta descriptions for a list of pages:
1. Fetch each page URL.
2. Extract the page title, H1, and first 300 words
   of visible body text.
3. Identify the primary topic and target keyword.
4. Write a meta description that:
   - Starts with the primary keyword or its close variant
   - Summarises what the user will get from the page
   - Is between 140 and 158 characters
   - Ends with a light call-to-action or benefit phrase
   - Does NOT include the site name (it appears separately)
   - Does NOT use generic phrases like "Learn more" alone
   - Reads naturally — not keyword-stuffed
5. Output a table: URL | Current meta | Suggested meta | Char count
6. Flag any page where the current meta is already good
   (skip generating a new one).

## schema-generator

When asked to generate FAQPage schema for a page:
1. Fetch the URL.
2. Find all question-and-answer pairs on the page.
   Look for: FAQ sections, H2/H3 headings phrased as
   questions, accordion items, definition lists.
3. For each pair, extract:
   - The exact question text (as the user wrote it)
   - A complete, self-contained answer (100–300 words)
     that makes sense without context from the question
4. Generate valid JSON-LD:

{
  "@context": "https://schema.org",
  "@type": "FAQPage",
  "mainEntity": [
    {
      "@type": "Question",
      "name": "[question text]",
      "acceptedAnswer": {
        "@type": "Answer",
        "text": "[answer text — HTML allowed but keep simple]"
      }
    }
  ]
}

5. Validate: every Question must have @type, name,
   and acceptedAnswer. Every Answer must have @type
   and text. Reject any pair where the answer is
   under 30 words.
6. Output the complete JSON-LD block, ready to paste
   inside a <script type="application/ld+json"> tag.

## schema-generator (Article mode)

When asked to generate Article schema for a blog post:
1. Fetch the URL and extract:
   - Page title (for "headline")
   - Meta description (for "description")
   - Publication date from the page or its URL slug
   - Author name (if present, else use site organisation)
   - First image URL (for "image", if present)
2. Generate valid JSON-LD Article schema including:
   - @type: "Article"
   - headline (max 110 chars)
   - description
   - url (canonical URL)
   - datePublished (ISO 8601 format: YYYY-MM-DD)
   - dateModified (same as published if unknown)
   - author (@type: Organization or Person)
   - publisher (Organization with name and url)
   - image (if available)
3. Also generate a BreadcrumbList based on the URL path:
   - home → [section] → [page title]
4. Output both schemas as separate JSON-LD blocks.

## schema-auditor

When asked to audit schema markup across a site:
1. Fetch sitemap.xml and get all page URLs.
2. For each page, fetch the HTML and find all
   <script type="application/ld+json"> blocks.
3. For each JSON-LD block, check:
   - Valid JSON (parse without error)
   - Required fields present for the @type:
     Article: headline, datePublished, author, publisher
     FAQPage: mainEntity with at least one Question
     Product: name, offers (with price and availability)
     BreadcrumbList: itemListElement with position, name, item
     LocalBusiness: name, address, telephone
   - No empty string values for required fields
   - datePublished in valid ISO 8601 format
   - URL fields are absolute (not relative) paths
4. Report:
   - Pages with no schema at all
   - Pages with schema errors (list field and issue)
   - Pages with valid schema (summary count)
5. Categorise: Critical (missing @type or all required
   fields), Warning (missing optional but recommended
   fields), Info (minor issues like relative URLs).

## Weekly New Page Schema Check — Every Wednesday 09:00

1. Check sitemap.xml for any pages added in the
   last 7 days.
2. For each new page, use schema-auditor to check
   whether valid schema is present.
3. For any new page missing schema, use
   schema-generator to draft the appropriate schema
   type based on the page content.
4. Send a summary to #seo-alerts on Slack:
   - New pages this week: [count]
   - Pages missing schema: [count + URLs]
   - Draft schema attached for review

## Monthly Schema Audit — 5th of each month, 08:00

Run schema-auditor on the full sitemap.
Generate a complete report of:
- Pages with no schema
- Pages with schema errors (grouped by error type)
- Pages with valid schema (count only)

For pages with critical errors, attempt to
regenerate the schema using schema-generator
and include the corrected version in the report.

Send the full report to [your email].
Save to ~/openclaw/reports/schema-audit-[date].md

## og-meta-checker

When asked to audit Open Graph tags:
For each page URL, check:
- og:title (present? Under 60 chars?)
- og:description (present? Under 200 chars?)
- og:image (present? Is it an absolute URL?
  Does the image URL actually return a 200 response?)
- og:type (present? Correct value for page type?)
- og:url (present? Matches canonical URL?)
- twitter:card (present?)

Flag any missing or incorrect tags.
If og:image is missing, note that sharing this
page will produce an ugly link preview.

For pages missing og:description, suggest one
based on the meta description (or generate a
new one if the meta description is also missing).