Zero-Knowledge Vault
Your passwords are encrypted on your device — our servers cannot read them. Unlock with Touch ID, autofill with anti-phishing domain matching, generate strong passwords. Available in the Chrome extension and web app.
AES-256-GCM encryption. Zero-knowledge vault. 5-layer tenant isolation. Zero-credential architecture. GDPR compliant. Your data stays yours.
Your passwords are encrypted on your device — our servers cannot read them. Unlock with Touch ID, autofill with anti-phishing domain matching, generate strong passwords. Available in the Chrome extension and web app.
Each agent runs inside two sealed containers that cannot see each other. One handles AI thinking, the other handles your tools. Architecturally impossible to cross boundaries.
Every container creates a unique identity in memory at boot. No stored keys, no shared secrets. Your API keys are injected at the network edge — invisible to the container.
Every marketplace skill passes through four layers of defense: pattern scan, AI analysis, community flags, and human review.
Every file gets its own unique encryption key. Chat messages and audit logs are encrypted at the database level. Credentials are zero-knowledge. Four independent layers — no single point of failure.
AES-256-GCM encryption at rest, TLS 1.3 in transit, row-level security on all tables, zero-knowledge vault. All data in GDPR-compliant EU data centers.
Yes. Operated by Christof Wallner (sole proprietor, Cyprus). Fully GDPR compliant. EU data centers. All data subject rights supported.
5-layer tenant isolation: row-level security on all database tables, container isolation per agent, zero-credential architecture, CSP strict-dynamic headers, and rate limiting across all edge functions.
No. Agent Magic never trains AI models on customer data. Sub-processors (Anthropic, OpenAI, Google) operate under strict DPAs prohibiting training on customer inputs.