OpenSurity v0.1 provides specific trust mechanisms designed for local development and direct decentralized handshakes. Understanding the boundaries of the prototype is critical.
- Tamper-evident behavioral history: All trust events are uniquely content-addressed and linked via an immutable SHA-256 chain.
- Cryptographic agent identity: Identity is strictly enforced at all trust boundaries. Operations require HMAC-SHA256 signatures at Level 1, and W3C DID:key Ed25519 signatures at Level 2.
- Replay attack prevention: The 5-step trust handshake utilizes a non-deterministic TTL nonce verification system stored locally.
- Key isolation: Cryptographic keys are isolated dynamically, stored on disk with
0o600permissions, and are never exposed inside manifests or state representations.
- End-to-end encryption of task payloads: We assume TLS is handled securely at the transport layer by the application environment.
- Zero-knowledge proofs: L3 identity (ZK-SNARK proofs) is designated for the future roadmap.
- Post-quantum signature security: Algorithmic upgrades like ML-DSA-65 are part of the future roadmap.
- Decentralized key revocation: There is no CRL or OCSP-equivalent revocation mechanism implemented in v0.
- Protection against a compromised agent runtime environment: OpenSurity intercepts communication at the framework level to establish trust, but agents inherently still execute arbitrary native code. Sandboxing is left strictly to the host environment.