Security

Security is the reason Daevix exists - autonomous agents are powerful and untrusted, and the platform’s job is to contain them. These two documents describe how.

Security Model

The security model defines Daevix’s security posture - what the platform protects against, where the trust boundaries lie, and what controls are in place. It covers the “never trust the agent” principle, the layers of defense (network, host, platform, and agent identity), and the cryptographic properties of the token and secrets systems.

Read this to understand what the platform guarantees and why.

Threat Model

The threat model takes the attacker’s perspective. It identifies threat actors, maps attack vectors against each trust boundary, evaluates how well existing controls hold up, and documents residual risks.

Read this to understand what can go wrong and how the platform responds.