The Agent Skills Directory

[SAFE]: The skill provides a structured framework for auditing Google Ads campaigns. It utilizes standard Google Ads Query Language (GAQL) statements to pull performance metrics. No unauthorized data exfiltration, obfuscation, or remote code execution patterns were found. The skill follows best practices by instructing the agent to differentiate between direct evidence and inference.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data from the Google Ads API and manual user exports, which constitutes a potential attack surface for indirect prompt injection. However, this is inherent to its data analysis function and no exploitation was observed.
Ingestion points: External data is retrieved via the google-ads-mcp search tool and through manual performance exports provided by the user in the chat.
Boundary markers: The skill does not explicitly define delimiters or specific 'ignore instructions' warnings for the processed data, which is standard for analysis-oriented skills.
Capability inventory: The skill uses the search tool from the Google Ads MCP and has permissions to read and write markdown files within the workspace/ads/ directory.
Sanitization: There are no explicit sanitization steps mentioned for the external data before it is processed by the agent.

google-ads-pmax