AI-native

Authorization rules by conversation

Describe who should access what; the AuthZ agent drafts an OpenFGA model and per-route rules, validates them, and dry-runs against your traffic before you enable.

  • Drafts who-can-do-what rules, route by route
  • Tests them against your real traffic first
  • Starts watch-only — enable when the report looks right
  • Guardrails against prompt injection
npx apiblaze agent authz
$ npx apiblaze agent authz acme
you › only owners read their orders
agent › drafted 1 route rule
  /publish → live, watch-only (reports, blocks nothing)
  /enable  → enforcing

# in a hurry? one shot, no conversation:
$ npx apiblaze rule "only owners read their orders" acme

Policy design without the PhD

Intent → model

Plain-English access intent becomes an OpenFGA model + rules.

Dry-run

Simulate against real 2xx/4xx traffic before enforcing.

Guardrailed

Prompt-injection defenses and a credit pre-gate keep it safe.

Try it in 30 seconds

No signup required. Claim it to your account whenever you’re ready.