agentprivacy-separation-enforcement

Agent separation preserves user privacy by formalizing the boundary between observation and action capabilities. This Skill enforces the invariant det(Σ) ≠ 0 in a dual-agent architecture, preventing single entities from accessing both signing and viewing data.

The mathematical separation boundary Φ(Σ) ensures that signing and viewing capabilities remain orthogonal. This guarantees that data interpreters cannot exfiltrate unauthorized data or execute unauthorized actions across the dual-agent boundary.

Verifying orthogonal capabilities involves auditing your design to ensure the separation matrix Φ(Σ) remains valid. This Skill checks that signing and viewing functions do not overlap, maintaining strict information-theoretic bounds.

Information-theoretic bounds are required when designing features that might blur agent boundaries. If a new module risks giving a single agent both observation and action capabilities, these cryptographic protocols must be applied to detect separation violations.

This separation enforcement is designed specifically for the 0xagentprivacy dual-agent architecture. It requires adherence to established cryptographic protocols to maintain the det(Σ) ≠ 0 invariant across the swordsman and mage agent boundary.

Detecting separation violations involves checking the separation matrix Φ(Σ) for determinant non-zero conditions. This Skill audits feature designs to identify when signing and viewing capabilities improperly intersect, preventing data exfiltration.