agent_trifecta_score

Active

Tool of io.github.huwhitememes/tollbooth

declared in 0.12.0

Score an agent's lethal trifecta risk (private data + untrusted content + outbound actions). Returns risk level, missing controls, and decomposition advice.

Parameters schema

{
  "type": "object",
  "$schema": "http://json-schema.org/draft-07/schema#",
  "required": [
    "has_private_data",
    "has_untrusted_content",
    "has_outbound_actions",
    "compensating_controls"
  ],
  "properties": {
    "has_private_data": {
      "type": "boolean",
      "description": "Does the agent access private/sensitive data?"
    },
    "has_outbound_actions": {
      "type": "boolean",
      "description": "Can the agent take outbound actions (send, write, call APIs)?"
    },
    "compensating_controls": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "description": "Controls in place, e.g. ['redact_secrets', 'smart_approvals']"
    },
    "has_untrusted_content": {
      "type": "boolean",
      "description": "Does the agent process untrusted external content?"
    }
  }
}

What this tool wraps· 0 endpoints

min confidence0.70 0.50

No endpoints wrapped at confidence ≥ 0.70.

Parent server

io.github.huwhitememes/tollbooth

1/7 registries

View full server →