Share LLM-generated content as a web page

Turn any LLM's output into a public web page with a single POST. You get a clean shareable link with a proper preview card — a better way to send model output than a wall of pasted text or a screenshot.

Publish it

Send the model's answer as the markdown content field, plus a title for the headline and link preview:

curl -X POST https://quicky.page/api/v1/publish \
  -H 'content-type: application/json' \
  -d '{
    "title": "What we learned from the Q3 data",
    "content": "## Top findings\n\n- Finding one\n- Finding two\n\n> [!NOTE]\n> The model flagged this as the highest-confidence takeaway."
  }'

The response gives you the public url and a private edit link. That URL is everything you need for the three ways people actually share content.

Three ways to share the result

1. As a link

Paste the url anywhere. The page is responsive and reads cleanly on desktop and mobile:

https://quicky.page/abc123

2. As a rich preview card

Every page ships Open Graph and Twitter Card tags built from its title and first lines, so dropping the link into Slack, iMessage, Discord, or a social post unfurls into a titled preview card — not a bare URL. Give the page a descriptive title and the card writes itself.

3. As an embed

Because the result is a normal web page at a stable URL, you can iframe it into a doc, a notebook, or an internal dashboard, or link to it from a button in another page. There's nothing proprietary to render — it's just HTML at a URL.

Make model output read well

Ask the model to return markdown — headings, lists, **bold**, and links all render natively.
Use > [!NOTE], > [!TIP], or > [!WARNING] blockquotes for callouts that draw the eye to the key point.
Fence code with triple backticks for syntax-highlighted code blocks.
For images, embeds (YouTube, Figma, CodePen), or buttons, pass an explicit qp.v1 blocks array instead of markdown.

Let the model publish it itself

Instead of piping output through your own code, you can give the model the publishing tool directly. The remote MCP server works in Claude, Cursor, VS Code, and ChatGPT Developer Mode; ChatGPT on other plans can use a Custom GPT Action. Then "publish this and give me the link" is a single instruction.