Overview
How Securix works - at a glance
Securix sits between your AI agent and external services, handling authentication, security, and monitoring so you don't have to.
High-Level Flow
Your AI Agent → Securix Proxy → Google/Gmail/Drive/etc
↓
[Token Vault]
↓
[Security Rules]
↓
[User Dashboard]What Securix Handles
- Authentication - OAuth with providers (Google, Microsoft, etc.)
- Token Storage - Vaults raw tokens securely
- Proxying - Forwards API requests with tokens injected
- MCP Gateway - Provides a secure Model Context Protocol interface for AI Agents
- Security - Applies user-defined rules and monitors for anomalies
- Monitoring - Provides dashboard for users to see agent activity
What You Handle
- Building your AI Agent - Your core logic
- Accessing User Data - Using your
securix-api-key,securix-agent-idand the user'ssecurix-entity-id - Displaying Data - Showing results to users
Traditional OAuth vs Securix
| Aspect | Traditional OAuth | Securix |
|---|---|---|
| Token Storage | Your database (high risk) | Securix Vault (encrypted) |
| User Control | All-or-nothing revoke | Granular (folder hide, read-only) |
| Monitoring | None | Real-time dashboard |
| Audit Costs | $15K+/year (you pay) | Included |
| Verification | You need to verify | Already verified |
Key Terminology
| Term | Definition |
|---|---|
| AASB | Agent Access Security Broker - the architectural pattern |
| entityId | A unique identifier for the user's connection (the "Entity") |
| Token Vault | SecuriX's encrypted storage for provider tokens |
| Virtual Scopes | User-defined permission overrides |
| Proxy Layer | Intercepts requests, injects tokens, applies rules |