skills/5dive-ai/skills/ad-creative/Gen Agent Trust Hub

ad-creative

Pass

Audited by Gen Agent Trust Hub on Jul 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Documents the use of local CLI tools (e.g., google-ads.js) and the Remotion framework to manage ad performance data and render video creative.
  • [EXTERNAL_DOWNLOADS]: References official repositories and packages for specialized tools such as Remotion (via npm) and Voicebox (via GitHub) for creative generation tasks.
  • [DATA_EXFILTRATION]: Provides implementation examples for transmitting creative briefs and scripts to well-known AI service providers, including Google, OpenAI, and ElevenLabs, via their official APIs.
  • [PROMPT_INJECTION]: Identifies an attack surface for indirect prompt injection through the ingestion of external performance data (CSV or API outputs) used to iterate on ad copy.
  • Ingestion points: Performance data provided by the user (CSV, paste, or API output) as described in SKILL.md.
  • Capability inventory: The agent can execute CLI tools and call network APIs across all documented workflows.
  • Boundary markers: The skill suggests a structured logic flow but does not specify explicit delimiters for untrusted data in the prompt structure.
  • Sanitization: No explicit mention of input validation or sanitization for external performance data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 5, 2026, 06:48 AM
Security Audit — agent-trust-hub — ad-creative