Smart Agent Security Model in AXQA
Security is at the core of Smart Agent’s design. It was built to allow powerful local execution without compromising infrastructure, permissions, or data isolation.
As part of the AXQA Execution Intelligence Platform, Smart Agent operates under a secure, hybrid execution architecture — enforcing authentication, permission validation, encrypted communication, and strict workspace isolation.
Security by design
- No inbound connections into your private network are required.
- The Agent initiates communication — not the other way around.
- All execution requests are authenticated.
- Workspace isolation is preserved at all times.
Authentication & authorization
Smart Agent requires secure authentication before it can retrieve or execute any test case.
- Each Agent session is tied to a specific user account.
- Permissions are validated before execution begins.
- Users can only execute test cases assigned to them.
- Project-level access rules are strictly enforced.
Note
The Smart Agent follows a zero-trust communication model — every execution request is authenticated and validated before processing.
Workspace isolation
- Each workspace operates independently.
- The Agent can only access data within its authenticated workspace.
- No cross-workspace visibility is possible.
Note
This ensures complete separation between organizations using the platform.
Execution integrity
- Execution source is clearly recorded (Smart Agent).
- Historical execution records cannot be modified after completion.
- Validation results are stored exactly as returned at runtime.
Secure communication
- All communication between the Agent and AXQA is encrypted.
- No sensitive credentials are exposed publicly.
- Token-based authentication prevents unauthorized access.
Internal infrastructure protection
- Internal APIs remain inside your network.
- No need to open firewall ports for external testing.
- The Agent executes locally and returns structured results only.
Note
This architecture aligns with strict corporate security policies by eliminating inbound exposure, preserving infrastructure isolation, and maintaining full audit traceability.
Permission enforcement
Even if an Agent is installed, it cannot bypass project rules.
- Execution permissions are validated server-side.
- Unauthorized execution attempts are denied.
- All actions remain traceable.
Best practices
- Install Smart Agent on controlled and secured machines.
- Limit local access to authorized testers.
- Regularly review assigned test cases and permissions.
- Use separate Agents for different environments when required.
Common mistakes
❌ Sharing Agent login credentials across multiple users
✔ Each tester should authenticate individually.
❌ Assuming local execution bypasses platform permissions
✔ All execution requests are validated against project access rules.
Security philosophy
Smart Agent extends execution capabilities without extending risk. It keeps systems internal, permissions enforced, and execution fully traceable — delivering flexibility without sacrificing control.
Related documentation
- Smart Agent Overview
- Internal Network Testing (High-Security Mode)
- Executing Test Cases via Smart Agent