Connect an agent
Install the SDK, create a scoped key, register tools, and confirm the first policy decision.
Open quickstart →Start with a verified setup path, diagnose common integration problems, or send a redacted support request to the Quasentra team.
Install the SDK, create a scoped key, register tools, and confirm the first policy decision.
Open quickstart →Assign ALLOW, DENY, or REQUIRE_APPROVAL and keep newly discovered tools closed.
Review permissions →Understand exact-operation approvals, expiration, argument binding, and one-time consumption.
Use approvals →Move supported HTTPS, PostgreSQL, or email credentials behind the trusted gateway.
Configure gateway →Use an execution witness and adversarial matrix to verify blocked tools never run.
Test the boundary →Check keys, versions, timeouts, failure behavior, credentials, and incident response.
Production checklist →Do not catch a security or transport error and call the original tool. Use the reason code and request ID to identify the actual configuration or policy issue.
Confirm the application ran with an API key containing agent:register, the base URL is correct, and startup did not suppress a registration error. Then review the registration checks.
Verify the normalized action name, agent ID, active policy, API-key scope, and tenant. Unknown or mismatched tools are denied by design.
The tenant, agent, action, resource, and canonical arguments must match the original request. Expired, denied, changed, or consumed approvals fail closed.
Validation checks stored configuration and encrypted credential integrity; it does not contact the remote service. Run a non-production staging call to verify DNS, TLS, login, and firewall access.
Supported adapters fail closed after bounded retries. Preserve the request ID, inspect service health, and never run the sensitive tool as a fallback.
Email is recommended for technical cases because it preserves request IDs, versions, and redacted evidence.
Email support →View all contact options