Prompts to Try with OpenGraph.io MCP
Copy-paste prompts for previewing URLs, extracting content, capturing screenshots, checking metadata, auditing sites, and generating images — organized by what you're trying to get done.
Every prompt below is phrased the way you'd naturally ask — no tool names required. Your MCP client picks the right one automatically.
Prompts by Category
Swap in your own URLs and paste directly into Cursor, Claude, or any other connected MCP client.
Preview URLs
“Preview https://example.com and summarize the title, description, image, and metadata.”
“Check how https://example.com/blog/latest-post looks when shared on Facebook, X, and LinkedIn.”
“Compare the Open Graph tags on these three URLs and tell me which is best optimized: url-1, url-2, url-3.”
Extract Content
“Convert https://example.com/docs/getting-started to Markdown and summarize the key steps.”
“Scrape https://example.com/pricing and list every plan name and price in a table.”
“Read https://example.com/refund-policy and tell me exactly what their refund window is.”
Screenshots
“Take a full-page screenshot of https://example.com and describe the layout.”
“Screenshot https://example.com in dark mode at a mobile viewport size.”
“Screenshot these two competitor homepages and compare their hero sections: url-1, url-2.”
Metadata & Social Preview
“Check if https://example.com has a valid og:image and tell me its dimensions.”
“Audit the Twitter Card tags on https://example.com and tell me what's missing.”
“Get the metadata for my last five blog posts and check they all have unique titles: url-1, url-2, url-3, url-4, url-5.”
Audit
“Run an instant audit on https://example.com/launch and give me the top three fixes.”
“Discover all the URLs on example.com, then start a full site audit.”
“Once the audit finishes, summarize the report and rank the issues by priority.”
“What changed on example.com since the last audit, and has anything we already fixed come back?”
“Audit example.com every week on Monday morning and tell me when the score drops.”
“For the pages that failed the title check, draft better titles, save them to the fix list, and export it as CSV.”
Research & Content
“Pull the three most recent articles from https://example.com/blog and summarize each in two sentences.”
“Compare how these two competitor pricing pages describe their enterprise plan: url-1, url-2.”
“Read https://example.com/docs/changelog and write a one-paragraph summary for release notes.”
Developer
“Scrape the API reference at https://example.com/api-docs and generate a TypeScript interface for the response shape.”
“Query https://example.com/docs/auth and tell me what authentication method their API uses.”
“Generate a simple architecture diagram showing how a client, an MCP server, and an API connect.”
Agency & Marketing
“Check the link preview for our new landing page before we launch it: https://example.com/launch.”
“Audit our client's homepage at https://example.com and give me three talking points for our next call.”
“Generate a square social card with the headline "Now Live" for our announcement post.”
Get Better Results From Your Prompts
- Give the agent a real URL: Paste the actual URL, not a description of one — the agent can't guess a domain from context alone.
- State the output format: "As a table," "in JSON," or "in two sentences" tells the agent exactly how to shape the answer.
- Chain steps in one prompt: "Preview this, then screenshot it" lets the agent call both tools without you re-prompting.
- Ask for a fix list, not just a score: For audits, request the top N issues so the agent prioritizes the results for you.
- Name the platform for previews: "On Facebook and LinkedIn" is more precise than "everywhere" and returns more specific feedback.
Use These Capabilities Directly, Too
Every prompt above is backed by a product you can also call directly from the REST API.
Trusted by Thousands
Over 1B URLs processed
Powering previews, metadata, screenshots, extraction, embeds, and audits.
