Configure tenant guardrails

Guardrail capability

The underlying control, such as prompt-injection protection, PII masking or topic restriction.

Guardrail policy

The tenant-specific configuration that selects a capability, optionally limits it to an environment and supplies any capability-specific settings.

Tenant interaction
        ↓
Matching environment policy
        ↓
Guardrail capability
        ↓
Monitor or enforce
        ↓
Status and violation indicators

Active guardrails

Policies currently active for the tenant.

Enforcing

Policies operating in enforcement mode rather than monitoring-only mode.

Violations (7d)

The number of recorded policy violations during the previous seven days, when violation data is available.

Availability

The platform-reported availability status for the selected guardrail capability.

Monitor

Observes or records matching activity without applying the full enforcement behaviour, where supported.

Enforce

Actively applies the configured policy behaviour.

Enabled

Controls whether the individual policy is switched on.

• Open the correct customer tenant.
• Decide whether the policy applies to every environment or one named environment.
• Identify the guardrail capability required by the customer use case.
• Prepare any blocked words, additional masked words or allowed topics.
• For a custom guardrail, obtain the exact approved capability ID and configuration schema.
• Decide whether the policy should initially monitor or enforce.
• Prepare representative test inputs, including acceptable content and content that should be flagged.

Enlarge

Active guardrails

Shows the number of active policies for the tenant.

Enforcing

Shows how many policies are operating in enforcement mode.

Violations (7d)

Shows the recent violation count when data is available.

Status label

The label beside the page title summarises whether the tenant currently has active guardrails.

The overview screenshot above represents a tenant with no configured guardrails. It is an example of the interface, not an example of a configured policy set.

1. 1Open the customer tenant.
2. 2Select Guardrails.
3. 3Select Add guardrail.
4. 4Choose the required capability.
5. 5Enter an environment when the policy should apply only to that environment.
6. 6Leave Environment blank when the policy should apply to all environments.
7. 7Complete the capability-specific settings.
8. 8Select Save.
9. 9Find the saved policy in the table.
10. 10Review its Availability, Mode and Enabled columns.
11. 11Test the policy with representative inputs.
12. 12Move to enforcement only after confirming that expected interactions continue to work.

Capability ID

The exact identifier of the approved guardrail capability, for example one supplied by the platform administrator or capability owner.

Environment

Optional environment restriction. Leave blank to apply the policy across environments.

Config (raw JSON)

The capability-specific configuration object. It must be valid JSON and conform to the schema expected by the capability.

1. 1Confirm that the capability is approved and available on the deployment.
2. 2Obtain the exact Capability ID.
3. 3Obtain the supported configuration schema and a safe example.
4. 4Select Custom from the Capability list.
5. 5Enter the Capability ID exactly.
6. 6Add an environment only when required.
7. 7Enter valid JSON in Config.
8. 8Review the JSON for syntax errors.
9. 9Select Save.
10. 10Confirm the saved policy's Availability status.
11. 11Test it in a controlled environment before enforcement.

1. 1Open Guardrails for the tenant.
2. 2Clear the search field.
3. 3Select the required environment or All environments.
4. 4Select All, Enforce or Monitor.
5. 5Review the Availability status.
6. 6Confirm the expected Mode.
7. 7Confirm that Enabled has the intended value.
8. 8Review the Updated date after making a change.
9. 9Use the row actions available in the current deployment to inspect or change the policy.
10. 10Test the affected tenant interaction after changing enforcement.

1. 1Open the correct tenant.
2. 2Select Guardrails.
3. 3Review the policies and environment filters.
4. 4Select Export policies.
5. 5Save the downloaded export in an approved customer or partner location.
6. 6Record the tenant and export date when the file will be used for governance review.

1. 1Agree the customer's governance objective.
2. 2Select the smallest relevant capability.
3. 3Target a non-production environment first where one is available.
4. 4Configure only the required words, phrases or topics.
5. 5Save the policy.
6. 6Confirm capability availability.
7. 7Enable Monitor mode where supported.
8. 8Test acceptable, borderline and unacceptable inputs.
9. 9Refine the configuration to reduce false positives.
10. 10Enable enforcement when the customer approves the behaviour.
11. 11Review the seven-day violation indicator.
12. 12Export the policy configuration for governance records.

Enlarge

1. 1Confirm that the administrator opened the intended tenant.
2. 2Confirm that the account has partnership administrator access.
3. 3Refresh the Guardrails page.
4. 4Return to the tenant overview and reopen Guardrails.
5. 5Check whether other tenant administration pages load normally.
6. 6Do not add or change policies while the current policy state cannot be retrieved.
7. 7Contact the Greentic platform administrator when the forbidden or unavailable state persists, including the tenant identifier, time of the error and a screenshot without sensitive data.