Skip to main content
Sign in →
SaaS

Policies in Dashboard

Create and manage allow, deny, and transform rules for your agents directly in the ShieldAgent dashboard — no YAML files, no deployments.

Policy types

Allow

Explicitly permit a tool call. Use to create an allowlist for sensitive tools.

Deny

Block a tool call and return an error to the agent. Logs the denial in the audit trail.

Transform

Modify the call arguments before forwarding (e.g. redact PII, strip credentials from output).

Creating a policy

  1. 1.Go to Policies → New Policy.
  2. 2.Select the tenant and optionally scope to a specific agent .
  3. 3.Choose the action: Allow, Deny, or Transform.
  4. 4.Define the match conditions(tool name, argument pattern, risk score range).
  5. 5.Set priority. Lower numbers run first. Conflicting rules: lowest priority wins.
  6. 6.Click Save & Activate. Policy is live immediately.

Example: block execute_command

ActionDENY
Toolexecute_command
ScopeAll agents
Priority1
ReasonPrevent arbitrary command execution

Policy templates

Start from a pre-built template under Policies → Templates:

  • ▸Read-only filesystem — deny all write/delete tool calls
  • ▸PII redaction — transform outputs to strip email, phone, SSN patterns
  • ▸EU AI Act baseline — deny high-risk tools, log everything to audit trail
  • ▸Risk-scored block — auto-deny calls scoring above a configurable threshold

Compliance Reports →

Generate EU AI Act, ISO 42001, NIST AI RMF evidence from your audit trail

Policies in Dashboard