03-danh-gia-hieu-suat
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references official Meta/Facebook MCP integration endpoints (mcp.facebook.com) and recognized third-party connectors (pipeboard.co) to automate the retrieval of marketing data. These references are informative and consistent with the skill's stated purpose.
- [DATA_EXPOSURE]: The skill is designed to process sensitive business metrics including advertising spend, ROAS, and lead quality. This data handling is appropriate for a performance audit tool and occurs within the established agent context.
- [COMMAND_EXECUTION]: While the skill includes logic for modifying ad campaigns, it explicitly enforces a 'verify before write' policy (Rule 7), ensuring that any automated changes via MCP require user approval.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from advertising platforms. Given the structured nature of marketing metrics (integers, floats, status codes), the risk of executing malicious natural language instructions through these data feeds is negligible.
Audit Metadata