Security
We don't ask you to trust us.
Below is a summary of how SlothBox enforces its trust guarantees. The canonical documents — threat model, full crypto details, runbook — live in the repository so they version with the code.
Cryptography
XChaCha20-Poly1305 IETF for symmetric AEAD. BLAKE2b-256 for key fingerprints and whole-file integrity. Argon2id for password-protected shares. All primitives are direct calls into libsodium — no custom logic, no hand-rolled MAC, no key stretching outside vetted defaults.
Threat model
The system protects content confidentiality from the SlothBox operator and from network observers. It does not protect against an endpoint compromise (sender or recipient). Explicit non-goals are documented so you can decide whether the model fits your use case.
Verifiable architecture
Every container, every config, every cryptographic call lives in one repo under MIT. `docker compose up -d` brings the entire production stack online on your machine. v1.0 ships an offline `slothbox-verify` CLI you can audit independently.
Audit roadmap
The v0.2 line — internal review only, hardened against the URL-leak races called out in the v0.1 warning. v1.0 — independent cryptographer review + third-party application pen test, with reports published under /audits/. This milestone will not be softened to ship faster.
Reporting a vulnerability
Send vulnerability reports privately to the maintainer at philipsloth1@gmail.com or use the contact form at philipsloth.com/contact. The v0.2 line has no bug bounty; v1.0 will, scope-limited.