Skip to main content

​
Overview

While shopping may be intermediated by agents, it is important for both retailers – and consumers – to maintain a relationship: to be able to power loyalty programs, personalization, preferences and more. The rMCP identity framework provides multiple schemes for authentication and authorization, ensuring that:
  • Agents can act on behalf of users with proper authorization
  • Retailers can verify the legitimacy of requests
  • Users maintain control over their data and purchasing decisions

​
Identity Schemes

rMCP supports multiple identity schemes to accommodate different use cases and security requirements: Identity verification through email delegation provides a familiar and accessible authentication method: How it works:
  1. Agent provides an email address for the human shopper
  2. Retailer sends a magic link to that email address
  3. Agent accesses the email inbox and “clicks” the magic link
  4. Agent is now authenticated as that email identity
{
  "identity": {
    "type": "magic_link",
    "email": "[email protected]",
    "agent_id": "agent_123",
    "session_token": "temporary_session_token"
  }
}
Pros:
  • Already supported by most identity providers (Auth0, Okta, etc.)
  • Familiar user experience
  • No additional infrastructure required
  • Works with existing customer databases
Cons:
  • No inherent agent identity verification
  • Requires consumer to grant email inbox access to agent
  • Potential privacy concerns with email access
  • Session management complexity

​
2. Decentralized IDs (DIDs) with Verifiable Credentials (VCs)

A privacy-preserving approach using W3C standards for decentralized identity: How it works:
  1. Both agents and users have DIDs (decentralized identifiers)
  2. Users issue Verifiable Credentials to agents with specific permissions
  3. Agents present these VCs to retailers as proof of authorization
  4. Retailers verify the credentials cryptographically
{
  "identity": {
    "type": "did_vc",
    "agent_did": "did:web:agent.example.com",
    "user_did": "did:key:z6MkhaXgBZDvotDkL5257faiztiGiC2QtKLGpbnnEGta2doK",
    "verifiable_credential": {
      "type": ["VerifiableCredential", "ShoppingAuthorization"],
      "issuer": "did:key:z6MkhaXgBZDvotDkL5257faiztiGiC2QtKLGpbnnEGta2doK",
      "credentialSubject": {
        "id": "did:web:agent.example.com",
        "permissions": ["purchase", "view_orders"],
        "expires": "2024-12-31T23:59:59Z"
      }
    }
  }
}
Pros:
  • Strong cryptographic verification
  • Privacy-preserving (selective disclosure)
  • No central authority required
  • Clear agent and user identity separation
  • Granular permission control
  • Revocable credentials
Cons:
  • Requires DID infrastructure
  • Higher technical complexity
  • Less familiar to users
  • Ecosystem still maturing

​
Trust and Verification

For email-based authentication:
  • Email ownership verification through magic link delivery
  • Session tracking to associate agent actions with user sessions
  • Activity logs showing all actions taken by agents on behalf of users
  • Email notifications for important actions (purchases, account changes)

​
DID/VC Approach

For decentralized identity:
  • Cryptographic proof of both agent and user identities
  • Verifiable credentials that prove authorization without revealing unnecessary data
  • Credential revocation allowing users to instantly revoke agent permissions
  • Audit trail with cryptographically signed agent actions
Regardless of the identity scheme used, rMCP ensures:
  • Explicit consent for agent authorization
  • Clear visibility into granted permissions
  • Easy revocation mechanisms
  • Transaction approval workflows for high-value purchases

​
Implementation Guidelines

​
For Retailers

  1. Leverage existing auth providers (Auth0, Okta, etc.) that support magic links
  2. Add agent tracking to associate sessions with specific agents
  3. Implement rate limiting on magic link requests per email
  4. Provide session management APIs for agents to maintain authentication

​
DID/VC Implementation

  1. Set up DID resolution infrastructure or use existing services
  2. Implement VC verification logic following W3C standards
  3. Define credential schemas for shopping authorizations
  4. Build credential status checking for revocation support

​
For Agent Developers

  1. Request email access permission from users transparently
  2. Implement secure email handling to retrieve magic links
  3. Manage session lifecycle including refresh and expiration
  4. Provide clear UI showing which email identity is being used

​
Using DIDs/VCs

  1. Generate and manage agent DIDs securely
  2. Request VCs from users with minimal required permissions
  3. Present credentials appropriately to retailers
  4. Handle credential expiration and renewal gracefully

​
Security Considerations

  • Rate limit magic link requests to prevent email bombing
  • Short expiration times for magic links (e.g., 15 minutes)
  • One-time use tokens that expire after clicking
  • Secure email access protocols for agents
  • Session timeout policies for inactive agents

​
DID/VC Security

  • Private key management for DIDs using secure enclaves
  • Credential validation including signature and expiration checks
  • Revocation checking before accepting any credential
  • Minimal disclosure using selective disclosure features
  • Secure credential storage on agent and user devices

​
General Security Practices

  • All communications must use HTTPS/TLS
  • Implement comprehensive audit logging
  • Monitor for anomalous agent behavior patterns
  • Provide emergency revocation mechanisms
  • Regular security audits of identity infrastructure

​
Next Steps