cross-channel-performance

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill instructions define an analytical workflow for paid growth reports. It uses hypothetical tool calls (e.g., meta_ads_insights, asa_profitability) to retrieve advertising performance data.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: While the skill references sensitive file paths (e.g., app-ads-context.md) and identifiers (e.g., rc_key), these are used within the context of the agent's internal workspace for reporting purposes. Placeholders like <sk_xxx> and <proj_xxx> are used instead of hardcoded credentials, following best practices for secret management.
  • [COMMAND_EXECUTION]: There are no shell commands, subprocess spawns, or privilege escalation patterns present in the skill instructions.
  • [PROMPT_INJECTION]: The instructions do not contain attempts to override safety guidelines, bypass filters, or extract system prompts. The language is purely task-oriented for data analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 10:18 PM
Security Audit — agent-trust-hub — cross-channel-performance