Secure RPC

AZTP provides secure RPC (Remote Procedure Call) capabilities for authenticated agent-to-agent communication.

Understanding Secure RPC

In AZTP, secure RPC:

• Authenticates both caller and target agents
• Verifies trust domain relationships
• Enforces policy checks
• Provides audit trails

RPC Flow

When an agent makes a secure call:

1. Caller's identity is verified
2. Target's identity is verified
3. Policy is checked
4. Call is executed
5. Events are logged

Best Practices

1. Target Identification

   • Use consistent target IDs
   • Include environment in ID
   • Follow SPIFFE ID patterns

2. Action Naming

   • Use descriptive action names
   • Follow a consistent pattern
   • Match policy rules

3. Error Handling

   • Handle authentication failures
   • Handle policy denials
   • Handle network issues

Common Patterns

Basic RPC Call

// Make a secure call to another agent
const result = await securedAgent.secureCall(
  'finance-agent',    // Target agent ID
  'calculate-tax',    // Action to perform
  { amount: 1000 }    // Payload data
);

Environment-Aware Calls

// Development setup
const devResult = await securedAgent.secureCall(
  'dev/finance-agent',
  'calculate-tax',
  { amount: 1000 }
);
 

Error Handling Pattern

try {
  const result = await securedAgent.secureCall(
    'finance-agent',
    'calculate-tax',
    { amount: 1000 }
  );
} catch (error) {
  if (error.code === 'IDENTITY_ERROR') {
    // Handle identity verification failure
  } else if (error.code === 'POLICY_DENIED') {
    // Handle policy denial
  } else {
    // Handle other errors
  }
}