Make your HTML hard to read. Not impossible β nothing client-side is. This raises the cost of casual copying; it doesn't stop a determined reader. Now with Python and PHP targets, and emoji-dominant junk at every level.
Obfuscation is not encryption. The output is your source, compressed, encoded and wrapped in a decoder whose identifiers have been renamed. There is no secret β the key, where one is used, sits in the same file.
DevTools always wins. Whatever the browser renders, the Elements panel can show you. A decoder that runs in the browser can be followed in the browser.
Python and PHP are encoded, not renamed. For those targets the whole source becomes a compressed payload that the generated file decodes and executes. Your function and variable names inside that payload are not renamed β read the decompressed source and they are all there. Only the loader around it is obfuscated. Python uses exec, PHP uses eval; both are the honest description of what the file does.
The decoder can be blocked. A Content-Security-Policy that forbids inline scripts will stop a generated HTML page from decoding itself, and the page will render blank. That is a site-owner decision, not a defect.
The copyright notice is the only legally meaningful remedy. Every generated file carries a readable notice naming the author and year, outside the obfuscated payload.
The preview iframe is granted allow-same-origin. document.open() needs it to work inside a sandbox, so only preview HTML you wrote yourself.
The preview renders the obfuscated output. If it stays blank, open it in a new tab β some browsers refuse document.open() inside a sandbox.