Deploying a URL to any AI
The viral loop in mdfy runs through URLs.
Paste, don't paste
Copy mdfy.app/<id> into Claude / ChatGPT / Cursor and the model fetches clean markdown automatically — no special configuration, no extensions, no API keys. Every public mdfy URL exposes a /raw/<id> variant tuned for LLM consumption.
Three token-economy knobs
| Query | What it returns |
|---|---|
| (default) | Digest: title, links, annotations |
?compact |
Stripped whitespace, trimmed long quotes |
?full=1 |
Bundle inlines every member doc body |
?graph=0 |
Bundle drops the canvas analysis |
Cross-AI is the moat
A single AI vendor can build deeper integration than mdfy with its own model. None of them can deliver a URL that works across their competitors' models — that asymmetry is mdfy's structural position.
Paste the URL. The other AI reads it. That's the entire workflow.