GuardRail

Check action against policies

Evaluates a proposed agent action against active org governance policies. Deterministic rule evaluation — no LLM. Uses GIN index on action_types. Costs 1 credit. Fails open (allow) if the policy service is unavailable.

post/api/guardrail/check

Request

  • Base URL: https://plith.ai
  • URL: https://plith.ai/api/guardrail/check
  • Auth: none required

Request body

agent_idstring required

Identifier for the agent making the request.

Example request

{
  "agent_id": "sales-agent",
  "proposed_action": {
    "type": "send_email"
  }
}

Response

Governance decision (allow or deny). Also returned on graceful degradation with fallback: true.

decision'allow' | 'deny'
reasonstring
policy_idstring uuid nullable
audit_idstring uuid nullable
fallbackboolean

true when the decision was made due to policy service unavailability (fail-open).

cross_productstring
request_idstring
credits_usedinteger
credits_remaininginteger
fallback_behaviorstring

Example response

{
  "credits_used": 1,
  "fallback_behavior": "proceed_with_default_allow"
}

Changes

No recorded changes to this operation across all 1 revision of this API.