Audit the code

Everything that runs on your machine, and the servers it talks to, lives in one public repository. Read it in your browser, clone it, or build it yourself.

View the source on GitHub →

The parts that matter most for verifying the encryption claims:

• crates/crypto — PQXDH hybrid handshake (X25519 + ML-KEM-768), the message ratchet, sender keys, AEAD, and the wire format.
• crates/keystore — identity keys, at-rest sealing, and the keyserver client.
• crates/stego — the cover-text layer that carries ciphertext inside ordinary-looking chat.
• crates/ipc — how the pieces fit together: send, receive, whitelists, and burns.
• docs/THREAT_MODEL.md — what OSL protects against, and what it doesn't.

The code is fully open; the operational secrets are not, and never could be:

• Server secrets — the keyserver's admin token, Stripe keys, and the license-signing secret live only as deployment secrets, never in the repository.
• The release-signing key and deploy credentials used to build and publish installers.

None of these can decrypt your messages — the keyserver never sees plaintext or your keys. They're the keys to the project's infrastructure, not to your conversations.

End-to-end encryption you can't inspect is just a promise. Publishing the full client lets anyone confirm that messages are encrypted the way we say they are, that keys never leave your device, and that there's no backdoor. Reading the code is a guarantee that you can check — not a guarantee that the code is bug-free. OSL has not had a paid third-party security audit yet; that's a goal, not a claim.

Read the crypto code on GitHub, then confirm the binary you downloaded corresponds to it. Reproducible builds are planned. If you find something wrong with the cryptography, please contact OSLPrivacy@gmail.com — security reports are welcome and we'd rather hear about it from you than from someone else.