Skip to content

Enforcement

Sentrie is a deterministic policy decision engine. It evaluates structured inputs against rules and returns true, false, or unknown. That’s it.

The engine is pure and side-effect-free—no mutations, no API calls, no state changes. This makes it safe to embed deep in your stack: IAM systems, API gateways, control planes, multi-tenant services, anywhere you need fast, reliable authorization decisions.

Because enforcement lives outside the policy, the same rules work across different systems with different enforcement modes - whether you’re blocking requests, redacting data, or triggering step-up auth.

Different platforms enforce differently:

  • IAM layers allow, deny, or escalate auth flows
  • API gateways throttle, rate-limit, or block routes
  • Multi-tenant systems control features and per-org limits
  • Security layers trigger reviews or step-up verification
  • Backend services serve, redact, or transform data

Keeping the engine pure gives you three guarantees:

  • Determinism — same input = same output, always
  • Reproducibility — decisions are safe to replay and audit
  • Portability — runs identically everywhere

This is what makes Sentrie safe for real-time, latency-sensitive paths.

Call Sentrie’s HTTP endpoint with context. Get a decision. Enforce it.

Terminal window
curl -X POST https://sentrie.host:7529/decision/namespace/policy/rule \
-H "Content-Type: application/json" \
-d '{
"facts": {
"user": "...",
"org": "...",
"roles": ["..."],
"plan": "pro"
}
}'

What happens next is up to you:

  • Backends accept, reject, or redact responses
  • IAM systems allow, deny, or trigger step-up auth
  • Gateways block, throttle, or downgrade routes
  • Feature gates enable, disable, or shadow-test features
  • Entitlement systems enforce quotas or usage limits
  • Data layers redact fields before returning them

Separating decision from action makes Sentrie:

  • Safe for critical paths
  • Consistent across systems
  • Easy to test and replay
  • Independently scalable
  • Flexible for different enforcement modes