mcp-control-operator
Pass
Audited by Gen Agent Trust Hub on Feb 21, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION] (SAFE): No malicious patterns detected. The skill uses normative RFC2119 language and explicit execution sequences to prevent instruction override and ensure deterministic agent behavior.
- [CREDENTIALS_UNSAFE] (SAFE): Security guidelines and contracts strictly mandate the use of workspace secret references and prohibit the inclusion of plaintext credentials in payloads or parameters.
- [DATA_EXFILTRATION] (SAFE): All network interactions are confined to the user-configured workspace gateway endpoint; no unauthorized data transmission or suspicious outbound patterns were identified.
- [REMOTE_CODE_EXECUTION] (SAFE): The skill manages automation logic via a deterministic set of validated tools. It does not perform unverified script downloads or dynamic code execution outside the validated workflow management context.
- [COMMAND_EXECUTION] (SAFE): Destructive actions, such as workflow deletion, are governed by explicit confirmation policies and require a 'validate-before-mutation' workflow step.
Audit Metadata