POST /api/mcp · no API key

QR codes your AI assistant can manage

Connect Claude, Cursor or any MCP client and it can make codes, change where they point after they're printed, and tell you how often they get scanned. Codes made this way don't expire either.

Claude Code
claude mcp add --transport http neverdie-qr https://neverdieqr.com/api/mcp
Cursor, Claude Desktop and other clients
{
  "mcpServers": {
    "neverdie-qr": {
      "url": "https://neverdieqr.com/api/mcp"
    }
  }
}
curl: make an editable code
curl -X POST https://neverdieqr.com/api/codes \
  -H 'content-type: application/json' \
  -d '{"destination":"https://example.com/menu","fallback":"https://example.com"}'
Plain image URL, no call needed
<img src="https://neverdieqr.com/api/qr?data=https%3A%2F%2Fexample.com" alt="QR code">

Tools

make_static_qr

Turns text or a URL into a QR image. Nothing on our side is involved after that, so it can't expire.

create_dynamic_qr

Makes an editable code with an optional fallback URL. Returns the scan URL to print and a manage token.

update_dynamic_qr

Points a printed code somewhere new, or sets or removes its fallback.

get_qr_stats

Total scans, scans per day for two weeks, recent scans and whether the destination is answering.

check_qr_code

Checks any NeverDie code without a token: is it live, where does it go, is the page up.

The token is the only key

There are no accounts. When an agent creates a code it gets a manage token back once. Store it with the code, because nobody can recover it later. If it's lost, you can no longer change where the code points or read its stats, but the printed code keeps scanning and going to the last destination you set.

Fallbacks keep scans landing

A check runs after a scan, at most every six hours, or every 15 minutes while the page is down. If the destination returns 404, 410 or a server error, or doesn't answer, scans go to the fallback until a later check finds it back. A code nobody scans doesn't get checked, and scans made before the first failed check still reach the broken page.

Limits

Each network can create 100 dynamic codes an hour, and that's meant to keep spammers from using the redirect. Static codes and image URLs have no limit.

Prefer plain HTTP? The same actions are in the OpenAPI spec, and llms.txt has the short version. Or make one by hand.