paid-channel-prioritizer

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to create and save a local markdown report (e.g., 'channel-strategy-[YYYY-MM-DD].md'). This is a legitimate functional requirement for exporting the analysis results to the user.
  • [EXTERNAL_DOWNLOADS]: The skill utilizes web search to access publicly available competitor data from advertising libraries. These downloads are limited to informational content for research purposes and do not involve executable code.
  • [PROMPT_INJECTION]: The skill processes untrusted external data via web search, creating a surface for indirect prompt injection.
  • Ingestion points: Competitor ad data retrieved through web search in Phase 1B.
  • Boundary markers: Absent; the skill does not use specific delimiters to separate search results from instructions.
  • Capability inventory: The agent has the ability to write files to the local system.
  • Sanitization: No explicit content validation or sanitization is performed on the search results before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 10:48 AM