Mizan AgentGate API

Use the free static scanner for discovery, then integrate authenticated pre-execution decisions into your own agent workflow.

Endpoints

AgentCheck example

curl -X POST "$MIZAN_URL/v1/agent/check" \
  -H "Content-Type: application/json" \
  -d '{
    "agent":{"id":"support-reader"},
    "permissions":[{"tool":"crm","operations":["read"]}],
    "controls":{"deny_by_default":true,"replay_protection":true,
      "audit_log":true,"intent_binding":true,"evidence_binding":true,
      "human_approval":{"required_for":["crm:delete"]}}
  }'

AgentGate example

curl -X POST "$MIZAN_URL/v1/agent/evaluate" \
  -H "Authorization: Bearer $AGENTGATE_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "request_id":"req-001",
    "intent":"Read a support ticket",
    "agent":{"id":"support-agent","role":"support"},
    "delegation":{"user_id":"u-1","scopes":["crm:read"],
      "declared_intent":"Read a support ticket","enforce_exact_intent":true},
    "policy":{"contract":{"mode":"READ_ONLY","allowed_roles":["support"],
      "allowed_tools":["crm"],"allowed_operations":["read"],
      "allowed_resource_kinds":["Ticket"]},
      "require_human_approval_for":[],"require_evidence_for":[]},
    "actions":[{"id":"read-ticket","tool":"crm","operation":"read",
      "resource":{"kind":"Ticket","name":"T-1"},"parameters":{}}]
  }'

Decision semantics

API access