Visual Social Preview & Open Graph Auditor

See the result instead of reading raw tags. This Actor inspects public URLs, renders a visual preview sheet, scores metadata health and returns concrete fixes that developers or AI agents can apply.

What makes this Actor different

Metadata extraction alone does not show whether a card will look right. Each successful audit can generate a PNG contact sheet covering:

• Facebook
• X / Twitter
• LinkedIn
• Slack
• Discord
• WhatsApp
• Google search

The previews are deterministic models built from the live title, description, domain and share image. They do not use generative AI. A platform can still apply its own crop, fallback or cached value.

Results

Every dataset row includes:

• A weighted metadata score and pass, warning and failure counts
• HTTP status and redirect information
• Resolved title, description and share image
• Canonical and robots signals
• Share-image reachability, type, size, dimensions and aspect ratio
• Open Graph and Twitter Card checks
• Platform readiness
• Concrete remediation snippets for every warning or failure
• A link to the generated PNG preview stored with the run

Input

Provide one to 100 public URLs. Choose a preview contact sheet or one specific platform, and optionally disable preview rendering for a faster metadata-only run.

{
  "urls": [
    "https://example.com/",
    "https://example.com/article"
  ],
  "renderPreviews": true,
  "previewPlatform": "all",
  "concurrency": 4
}

AI agents and automation

The input and output schemas are designed for APIs, scheduled audits, no-code workflows and Apify MCP. Agents can inspect the structured checks, retrieve the PNG preview when visual reasoning is useful, and apply the returned fix plan.

Security and privacy

The Actor processes only URLs explicitly supplied in a run. Private, local and reserved network destinations are blocked, including unsafe redirect targets. Requests have bounded timeouts, redirect counts and response sizes. No browser history, cookies or credentials are collected.