ChronoLock: The Anti-SaaS Secret Link
Most “secure share” tools still touch your secret.
They encrypt on their servers, require accounts, log metadata, or retain the ability to decrypt later. ChronoLock does none of that.
It encrypts in the browser. The server only ever sees ciphertext. There is no account and no telemetry. When the lifetime ends or the secret is revealed once (if you choose), it is gone.
The full source is public: github.com/Nuem8/ChronoLock (GPL-3.0).
The core rule
Everything sensitive happens in the browser using the Web Crypto API.
A random 32-byte key and 12-byte IV are generated locally. The secret is encrypted with AES-GCM before anything is uploaded. The server receives only the ciphertext plus a few flags (TTL, burn-after-read, whether it is locked to a public key).
You can verify this in two ways:
- Open DevTools → Network while creating a secret. Only ciphertext leaves the browser.
- Read the source. The encryption path is in the web client; the server never sees key material.
Two ways to lock a secret
1. Key in the link (default)
The raw AES key is placed in the URL fragment (after #). Browsers never send the fragment to the server. Anyone with the complete link can decrypt; a truncated link cannot.
2. Locked to a recipient public key
You paste the recipient’s public key. The secret is encrypted so that only the matching private key can open it. The link itself contains no decryption material. The private key never leaves the recipient’s browser and is never sent to ChronoLock.
If the private key is lost, the secret is unrecoverable. That is intentional.
What the server actually stores
- Ciphertext
- Chosen lifetime (5 min / 1 h / 1 day / 7 days)
- Whether to delete after one successful reveal
- Whether the secret is locked to a public key
Nothing else. No plaintext, no key, no user identifier, no visit log that identifies the secret.
On reveal the browser downloads the ciphertext and decrypts it locally. If “Delete after one reveal” is enabled, the ciphertext is removed only after a successful decryption with the correct key. A wrong key does nothing. The row is also deleted when its lifetime ends.
The server can run with an in-memory store (dies with the process) or SQLite (keeps unexpired ciphertext across restarts, still without the key).
Limits and deliberate constraints
- Maximum payload: 32 KB (enough for typical
.envfiles and small configuration blocks) - You can load a local
.envor text file — it is read only in the browser. The original filename is preserved client-side so the recipient can save it with the same name. - Encryption only works over HTTPS or localhost
- No accounts, no key recovery, no admin override
- Longest lifetime is 7 days by default
These are not missing features. They are the design.
Run it yourself
bash
cargo run --release -- --listen 127.0.0.1:8787 --store sqlite --data chronolock.db
Top comments (2)
👌 "trust no one, not even the server" energy here—refreshing to see a security tool that actually treats zero-knowledge as a strict architectural constraint rather than a marketing buzzword!
Thanks — that’s exactly the bar I was aiming for.
Most tools in this space still keep a quiet path where the server could see the secret if it wanted to. ChronoLock is built so that path simply doesn’t exist. The key never leaves the browser, the server only ever holds ciphertext it cannot decrypt, and the source is open so anyone can verify the claim.
Appreciate you noticing the difference.