emurlin

This was motivated for situations where using GPG just isn't practical. The issue is that there are very limited options to share a file securely without needing to install additional software (ZIP files come close to this, with some important caveats).

As I was making this, I tried to mitigate many of the concerns raised in that article. For example:

> ask for an file that can be decrypted with an ordinary standalone decryption tool

You don't need to run the self-decryption code at all. You can use 'ordinary' tools like `openssl` to decrypt the contents (note to self, add the `openssl` instructions to the `<noscript>` rendering).

> People who publish decryption tools expend effort to ensure that recipients can trust the actual decryption tools themselves before running them.

You _can_ verify the integrity of the tool itself using gpg. Since you can't verify the entire HTML file (otherwise, you wouldn't be able to inject the encrypted payload), there's ways a malicious actor could add additional scripts and still pass the integrity check. However, if you're diligent enough to be using gpg to check this, you'd likely also be diligent enough to spot such additions, or you'd be using the openssl route.

ZIP files can use two types of encryption, ZipCrypto or AES. As the sibling comment points out, ZipCrypto is pretty broken. IIRC the AES encryption is OK or mostly OK, but it's not universally supported, so your recipient may or may not be able to open the file you send them. Notably, the Windows built-in ZIP implementation didn't support ZipCrypto (this may have changed now that Windows ships with bsdtar / libarchive).

Subjectively, I feel this can be simpler to use (mostly for the recipient). You can host it at some server, share the link and the password and the recipient can access it like a regular website and get a plaintext download file. Compare that with a ZIP file, which with the same flow give you a ZIP download: if they don't immediately extract it, then they may have to hunt for the password later.

The goal was:

1. Something that didn't require any installation (assuming a web browser)

2. Have a single file with no network that could self-decrypt

3. Be fully auditable

The second point is done by having (sort of(*)) reproducible builds and embedded OpenPGP signatures.

The first point is made by cleverly manipulating the HTML structure so that it can decrypt without breaking the PGP signature. It can even decrypt using bare openssl (which was a design goal too, though getting the exact structure right took some work and bug reports).

The third point is accomplished by the first two, and by the source being freely available.

(*) Depends on the OS at the moment.

The project is split into two parts, one which is this ready-to-use web UI (obviously, useful for testing in the browser), but also a minimal library that can orchestrate the tests and produce a report (with some nice CLI-based reporters too).