The Agent Skills Directory

[Data Access and Tooling]: The skill uses specific MCP (Model Context Protocol) tools (search, list_accessible_customers, get_resource_metadata) to interact with the Google Ads API. This is a controlled way to access account data, as the tools themselves are responsible for managing authentication and scoping access to the Google Ads platform.- [Explicit Security Constraints]: The skill contains a specific section ('Do Not Attempt Unauthorized Local Probing') that instructs the agent not to run shell commands, custom Python scripts, or local database queries. This is a positive security pattern that reinforces the agent's boundaries and prevents it from attempting risky operations if the primary tools are unavailable.- [Indirect Prompt Injection Surface]: The skill ingests data from Google Ads resources (such as campaign names or conversion action names). While this represents a surface for indirect prompt injection, the data is structured and processed for diagnostic purposes within the Google Ads domain, presenting a low risk typical of API-integration skills.- [Query Generation]: The skill provides templates for GAQL (Google Ads Query Language). These are used strictly within the context of the search tool and do not involve dynamic execution of arbitrary code.

google-ads-api-account-diagnostics