AI Security reference architecture
MCP security
Identity, authorization, inspection and audit for agents calling tools through Model Context Protocol servers.
Design principles
- Authenticate every client and server
- Authorize per tool, not per server
- Never put secrets in model context
- Record every tool call
Components
| Component | Role |
|---|---|
| MCP gateway | Single enforcement point |
| Identity provider | User and agent identity |
| Policy engine | Tool-level decisions |
| Secrets broker | Downstream credentials |
More AI Security architectures
AI security architecture
A security layer between applications and models covering prompts, data, agents, MCP and policy.
View architectureAI agent security
Scoped permissions, approvals and monitoring for autonomous and semi-autonomous agents.
View architectureLLM governance
Inventory, risk tiers, approved providers and review gates for enterprise use of language models.
View architectureAdapt this to your environment
We'll help you apply this pattern to your constraints and existing estate.

