dollar-consent-first-callback
Pass
Audited by Gen Agent Trust Hub on Aug 4, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [SAFE]: The skill is designed specifically to enhance security by providing a managed escalation path for blocked destructive actions without granting the agent autonomous permission to execute them. It follows best practices for human-in-the-loop verification and confirmation tokens.
- [NO_CODE]: The skill consists entirely of instructional markdown and schema definitions. It does not include Python scripts, Node.js packages, or binary executables, significantly reducing the surface area for traditional malware or hidden code execution.
- [PROMPT_INJECTION]: The skill processes external data such as action summaries and workspace labels, which represents an indirect prompt injection surface. This is mitigated by explicit instructions to redact project details and to strictly ignore any natural language instructions that might be contained within a call transcript or owner response.
- [DATA_EXFILTRATION]: The skill triggers an external phone call via the CALL-E service, which constitutes a network boundary crossing. The risk is mitigated by strict data boundary rules that prohibit sending credentials, absolute home directory paths, source code, or complete agent transcripts over the call.
Audit Metadata