When you share a link, platforms and crawlers do not read your whole page. They look for Open Graph (OG), Twitter Card, and sometimes JSON-LD metadata, then pick a title, description, and image for the preview card. Auditing that metadata across many URLs is tedious if you copy tags by hand.
Metadata Packer in CohesionStack Labs accepts up to ten public URLs per run, extracts the tags the tool can see, downloads images and favicons, and gives you a spreadsheet plus a ZIP you can hand to editors, SEO specialists, or campaign teams. This guide explains what a run produces, who it helps, and how to use the exports. For a single-link check with platform-style previews, use Link Preview instead.
What a run produces
Each completed run yields two primary artifacts:
- Run metadata (CSV): Open Graph, Twitter Card, and JSON-LD fields where present, plus fetch URLs, HTTP status, canonical and final URLs, media dimensions, and paths to saved assets.
- Run assets (ZIP): Downloaded preview images, favicons, and generated preview cards tied to that run.
Together these form a reviewable metadata pack suitable for audits, creative handoffs, or before/after checks after you fix tags on site.
Who benefits and when
Batch packaging is aimed at teams that need reliable, shareable evidence at scale:
- Content creators and editors: Audit titles, descriptions, and images across a set of links before publishing roundups or newsletters.
- SEO and AEO specialists: Spot missing or inconsistent OG, Twitter, or JSON-LD fields; export CSVs for reporting and remediation.
- Marketing and social teams: Assemble media packs with canonical images and preview assets for campaign briefs or creative approvals.
- Developers and infrastructure: Validate redirects and canonical tags in bulk, integrate with CI checks, or use the API for automated audits.
If you only need an occasional single-URL preview, Link Preview is usually enough. Metadata Packer shines when you have a list and need exports everyone can open offline.
Open Graph tags: a quick primer
Open Graph meta tags are small snippets in a page's HTML that suggest which title, description, and image to use when a link is shared. They help platforms show a consistent, attractive preview.
Common tags include og:title, og:description, og:image, and og:url. Twitter Card tags (twitter:card, twitter:title, and related fields) serve a similar role on X and other surfaces. JSON-LD structured data can add context for search and rich results, though not every platform uses it for link cards.
For deeper background, see The Open Graph Protocol, MDN on meta elements, CSS-Tricks on social meta tags, and our Master SEO: OpenGraph, Meta Tags, and AI Insights article.
How platforms build preview cards
When a URL is shared, social platforms and many crawlers read OG and Twitter metadata (and sometimes JSON-LD) to build the card: title, description, and image. Collecting those tags and the actual image bytes makes it easier to diagnose broken previews, wrong crops, or missing descriptions before a campaign goes live.
Strict vs relaxed mode
Metadata Packer offers two extraction modes:
- Strict: Reports tags found in the page without applying friendly fallbacks. Use this when you need an honest audit of what is declared in HTML.
- Relaxed: May apply documented fallbacks when primary tags are missing, closer to how some consumers infer a preview. Use this when you want a practical picture of what might render even if markup is incomplete.
Choose the mode before you start a run; live runs respect the selection shown in the tool.
Using the CSV and ZIP exports
CSV columns capture OG, Twitter, and JSON-LD fields, fetch URLs, HTTP status, and media dimensions. Filter in a spreadsheet or import into your CMS to find missing images, inconsistent titles, or broken links, then re-run after fixes to confirm changes.
The ZIP bundles the binary assets referenced in the CSV: images, favicons, and generated preview cards. Share the ZIP with designers or stakeholders who need files, not just tag text.
Run-relative asset paths
When an image or favicon is downloaded for a run, you may see a run-relative path (for example images/foo.png). In the Labs UI, open thumbnails to view the served asset. If an asset fails to load, work through the troubleshooting steps below before assuming the source site is wrong.
Troubleshooting missing images
If an image or favicon appears missing (404), try these steps in order:
- Refresh the page (Ctrl/Cmd+R).
- Open the thumbnail in a new tab to see the direct asset URL.
- Clear cache or use a private window to rule out stale browser cache.
- Try another browser or device on the same network.
- Confirm you are using the exact preview or run link you were given (localhost links only work on the hosting machine).
- If the asset still fails, note the run ID, copy the failing asset URL, and share both with whoever hosted the preview.
Try Metadata Packer
Ready to extract a pack? Open Metadata Packer, paste up to ten URLs (one per line), choose strict or relaxed mode, and run Extract and package. Expand each result row to inspect metadata and platform-style previews, then download the CSV and ZIP when the run finishes.
We welcome feedback in the CohesionStack Reddit community. Briefly describe the improvement, why it would help, and include example URLs or a screenshot when relevant.