Skip to main content

Security Hardening

Production CAR deployments require multiple layers of security controls to prevent token theft, identity spoofing, behavioral manipulation, and data exfiltration.

Sender-Constrained Tokens (DPoP)

CAR mandates DPoP (RFC 9449) tokens to prevent bearer token theft:

Agent → generates ephemeral key pair
Agent → creates DPoP proof JWT
Auth Server → binds access token to public key
Resource Server → verifies token + DPoP proof match

Token lifetimes are capped by trust tier:

TierMax Token Lifetime
T0–T25 minutes
T3–T430 minutes
T5–T61 hour
T74 hours

TEE Binding

Agents at T5+ MUST provide Trusted Execution Environment attestation:

PlatformAttestation
Intel SGXDCAP
AWS NitroNitro document
AMD SEV-SNPSEV report
ARM TrustZonePSA token

TEE binding ensures code integrity — the agent binary hasn't been tampered with.

Semantic Governance

Layer 3 controls prevent prompt injection and data exfiltration:

Instruction Integrity

Agents are bound to pre-approved instruction sets. Unapproved instructions are rejected.

Output Schema Binding

Agent outputs must conform to approved schemas. Data exfiltration patterns are blocked.

Context Authentication

Context sources are authenticated to prevent indirect prompt injection.

Dual-Channel Authorization

Control plane (config changes) and data plane (content processing) are separated.

Revocation

SLA by Tier

TierMax PropagationCheck Type
T5–T7≤ 1 secondSynchronous
T3–T4≤ 15 secondsPre-action
T1–T2≤ 60 secondsPeriodic
T0≤ 5 minutesBest effort

Recursive Revocation

When a delegator is revoked, all delegatees in the chain are automatically revoked.

Security Levels

LevelNameRequirements
SH-1BasicTLS, signed tokens, basic key management
SH-2StandardDPoP, pairwise DIDs, revocation checks
SH-3HardenedTEE binding, ≤1s revocation, delegation verification, full audit

Anti-Gaming Measures

  • Sudden score jumps: Trigger manual review
  • Oscillating behavior: Caps maximum achievable tier
  • Rapid context switching: Resets behavior scoring
  • Coordinated manipulation: Cross-agent correlation detection